diff --git a/docs/dyn/spanner_v1.projects.instances.databases.html b/docs/dyn/spanner_v1.projects.instances.databases.html index aa4b0dc431d..a3a1780ca3c 100644 --- a/docs/dyn/spanner_v1.projects.instances.databases.html +++ b/docs/dyn/spanner_v1.projects.instances.databases.html @@ -455,6 +455,7 @@

Method Details

"token": "A String", # The token identifying the message, e.g. 'METRIC_READ_CPU'. This should be unique within the service. }, "startKeyIndex": 42, # The index of the start key in indexed_keys. + "timeOffset": "A String", # The time offset. This is the time since the start of the time interval. "unit": { # A message representing a user-facing string whose value may need to be translated before being displayed. # The unit of the metric. This is an unstructured field and will be mapped as is to the user. "args": { # A map of arguments used when creating the localized message. Keys represent parameter names which may be used by the localized version when substituting dynamic values. "a_key": "A String", diff --git a/docs/dyn/spanner_v1.projects.instances.databases.sessions.html b/docs/dyn/spanner_v1.projects.instances.databases.sessions.html index 1cbdde8390f..a3e75975498 100644 --- a/docs/dyn/spanner_v1.projects.instances.databases.sessions.html +++ b/docs/dyn/spanner_v1.projects.instances.databases.sessions.html @@ -179,7 +179,7 @@

Method Details

The object takes the form of: { # The request for BeginTransaction. - "options": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Required. Options for the new transaction. + "options": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Required. Options for the new transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -310,7 +310,7 @@

Method Details

"transactionTag": "A String", # A tag used for statistics collection about this transaction. Both request_tag and transaction_tag can be specified for a read or query that belongs to a transaction. The value of transaction_tag should be the same for all requests belonging to the same transaction. If this request doesn’t belong to any transaction, transaction_tag will be ignored. Legal characters for `transaction_tag` values are all printable characters (ASCII 32 - 126) and the length of a transaction_tag is limited to 50 characters. Values that exceed this limit are truncated. }, "returnCommitStats": True or False, # If `true`, then statistics related to the transaction will be included in the CommitResponse. Default value is `false`. - "singleUseTransaction": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute mutations in a temporary transaction. Note that unlike commit of a previously-started transaction, commit with a temporary transaction is non-idempotent. That is, if the `CommitRequest` is sent to Cloud Spanner more than once (for instance, due to retries in the application, or in the transport library), it is possible that the mutations are executed more than once. If this is undesirable, use BeginTransaction and Commit instead. + "singleUseTransaction": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute mutations in a temporary transaction. Note that unlike commit of a previously-started transaction, commit with a temporary transaction is non-idempotent. That is, if the `CommitRequest` is sent to Cloud Spanner more than once (for instance, due to retries in the application, or in the transport library), it is possible that the mutations are executed more than once. If this is undesirable, use BeginTransaction and Commit instead. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -421,14 +421,7 @@

Method Details

"a_key": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. "code": "A String", # Required. The TypeCode for this type. - "structType": { # `StructType` defines the fields of a STRUCT type. # If code == STRUCT, then `struct_type` provides type information for the struct's fields. - "fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. - { # Message representing a single field of a struct. - "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. - }, - ], - }, + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. }, }, "params": { # Parameter names and values that bind to placeholders in the DML string. A parameter placeholder consists of the `@` character followed by the parameter name (for example, `@firstName`). Parameter names can contain letters, numbers, and underscores. Parameters can appear anywhere that a literal value is expected. The same parameter name can be used more than once, for example: `"WHERE id > @msg_id AND id < @msg_id + 100"` It is an error to execute a SQL statement with unbound parameters. @@ -438,7 +431,7 @@

Method Details

}, ], "transaction": { # This message is used to select the transaction in which a Read or ExecuteSql call runs. See TransactionOptions for more information about transactions. # Required. The transaction to use. Must be a read-write transaction. To protect against replays, single-use transactions are not supported. The caller must either supply an existing transaction ID or begin a new transaction. - "begin": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. + "begin": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -453,7 +446,7 @@

Method Details

}, }, "id": "A String", # Execute the read or SQL query in a previously-started transaction. - "singleUse": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. + "singleUse": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -486,7 +479,11 @@

Method Details

"fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. { # Message representing a single field of a struct. "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. + "type": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. # The type of the field. + "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. + "code": "A String", # Required. The TypeCode for this type. + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. + }, }, ], }, @@ -563,14 +560,7 @@

Method Details

"a_key": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. "code": "A String", # Required. The TypeCode for this type. - "structType": { # `StructType` defines the fields of a STRUCT type. # If code == STRUCT, then `struct_type` provides type information for the struct's fields. - "fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. - { # Message representing a single field of a struct. - "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. - }, - ], - }, + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. }, }, "params": { # Parameter names and values that bind to placeholders in the SQL string. A parameter placeholder consists of the `@` character followed by the parameter name (for example, `@firstName`). Parameter names must conform to the naming requirements of identifiers as specified at https://cloud.google.com/spanner/docs/lexical#identifiers. Parameters can appear anywhere that a literal value is expected. The same parameter name can be used more than once, for example: `"WHERE id > @msg_id AND id < @msg_id + 100"` It is an error to execute a SQL statement with unbound parameters. @@ -591,7 +581,7 @@

Method Details

"seqno": "A String", # A per-transaction sequence number used to identify this request. This field makes each request idempotent such that if the request is received multiple times, at most one will succeed. The sequence number must be monotonically increasing within the transaction. If a request arrives for the first time with an out-of-order sequence number, the transaction may be aborted. Replays of previously handled requests will yield the same response as the first execution. Required for DML statements. Ignored for queries. "sql": "A String", # Required. The SQL string. "transaction": { # This message is used to select the transaction in which a Read or ExecuteSql call runs. See TransactionOptions for more information about transactions. # The transaction to use. For queries, if none is provided, the default is a temporary read-only transaction with strong concurrency. Standard DML statements require a read-write transaction. To protect against replays, single-use transactions are not supported. The caller must either supply an existing transaction ID or begin a new transaction. Partitioned DML requires an existing Partitioned DML transaction ID. - "begin": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. + "begin": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -606,7 +596,7 @@

Method Details

}, }, "id": "A String", # Execute the read or SQL query in a previously-started transaction. - "singleUse": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. + "singleUse": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -637,7 +627,11 @@

Method Details

"fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. { # Message representing a single field of a struct. "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. + "type": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. # The type of the field. + "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. + "code": "A String", # Required. The TypeCode for this type. + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. + }, }, ], }, @@ -703,14 +697,7 @@

Method Details

"a_key": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. "code": "A String", # Required. The TypeCode for this type. - "structType": { # `StructType` defines the fields of a STRUCT type. # If code == STRUCT, then `struct_type` provides type information for the struct's fields. - "fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. - { # Message representing a single field of a struct. - "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. - }, - ], - }, + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. }, }, "params": { # Parameter names and values that bind to placeholders in the SQL string. A parameter placeholder consists of the `@` character followed by the parameter name (for example, `@firstName`). Parameter names must conform to the naming requirements of identifiers as specified at https://cloud.google.com/spanner/docs/lexical#identifiers. Parameters can appear anywhere that a literal value is expected. The same parameter name can be used more than once, for example: `"WHERE id > @msg_id AND id < @msg_id + 100"` It is an error to execute a SQL statement with unbound parameters. @@ -731,7 +718,7 @@

Method Details

"seqno": "A String", # A per-transaction sequence number used to identify this request. This field makes each request idempotent such that if the request is received multiple times, at most one will succeed. The sequence number must be monotonically increasing within the transaction. If a request arrives for the first time with an out-of-order sequence number, the transaction may be aborted. Replays of previously handled requests will yield the same response as the first execution. Required for DML statements. Ignored for queries. "sql": "A String", # Required. The SQL string. "transaction": { # This message is used to select the transaction in which a Read or ExecuteSql call runs. See TransactionOptions for more information about transactions. # The transaction to use. For queries, if none is provided, the default is a temporary read-only transaction with strong concurrency. Standard DML statements require a read-write transaction. To protect against replays, single-use transactions are not supported. The caller must either supply an existing transaction ID or begin a new transaction. Partitioned DML requires an existing Partitioned DML transaction ID. - "begin": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. + "begin": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -746,7 +733,7 @@

Method Details

}, }, "id": "A String", # Execute the read or SQL query in a previously-started transaction. - "singleUse": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. + "singleUse": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -778,7 +765,11 @@

Method Details

"fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. { # Message representing a single field of a struct. "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. + "type": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. # The type of the field. + "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. + "code": "A String", # Required. The TypeCode for this type. + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. + }, }, ], }, @@ -913,14 +904,7 @@

Method Details

"a_key": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. "code": "A String", # Required. The TypeCode for this type. - "structType": { # `StructType` defines the fields of a STRUCT type. # If code == STRUCT, then `struct_type` provides type information for the struct's fields. - "fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. - { # Message representing a single field of a struct. - "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. - }, - ], - }, + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. }, }, "params": { # Parameter names and values that bind to placeholders in the SQL string. A parameter placeholder consists of the `@` character followed by the parameter name (for example, `@firstName`). Parameter names can contain letters, numbers, and underscores. Parameters can appear anywhere that a literal value is expected. The same parameter name can be used more than once, for example: `"WHERE id > @msg_id AND id < @msg_id + 100"` It is an error to execute a SQL statement with unbound parameters. @@ -932,7 +916,7 @@

Method Details

}, "sql": "A String", # Required. The query request to generate partitions for. The request will fail if the query is not root partitionable. The query plan of a root partitionable query has a single distributed union operator. A distributed union operator conceptually divides one or more tables into multiple splits, remotely evaluates a subquery independently on each split, and then unions all results. This must not contain DML commands, such as INSERT, UPDATE, or DELETE. Use ExecuteStreamingSql with a PartitionedDml transaction for large, partition-friendly DML operations. "transaction": { # This message is used to select the transaction in which a Read or ExecuteSql call runs. See TransactionOptions for more information about transactions. # Read only snapshot transactions are supported, read/write and single use transactions are not. - "begin": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. + "begin": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -947,7 +931,7 @@

Method Details

}, }, "id": "A String", # Execute the read or SQL query in a previously-started transaction. - "singleUse": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. + "singleUse": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -1029,7 +1013,7 @@

Method Details

}, "table": "A String", # Required. The name of the table in the database to be read. "transaction": { # This message is used to select the transaction in which a Read or ExecuteSql call runs. See TransactionOptions for more information about transactions. # Read only snapshot transactions are supported, read/write and single use transactions are not. - "begin": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. + "begin": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -1044,7 +1028,7 @@

Method Details

}, }, "id": "A String", # Execute the read or SQL query in a previously-started transaction. - "singleUse": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. + "singleUse": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -1130,7 +1114,7 @@

Method Details

"resumeToken": "A String", # If this request is resuming a previously interrupted read, `resume_token` should be copied from the last PartialResultSet yielded before the interruption. Doing this enables the new read to resume where the last read left off. The rest of the request parameters must exactly match the request that yielded this token. "table": "A String", # Required. The name of the table in the database to be read. "transaction": { # This message is used to select the transaction in which a Read or ExecuteSql call runs. See TransactionOptions for more information about transactions. # The transaction to use. If none is provided, the default is a temporary read-only transaction with strong concurrency. - "begin": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. + "begin": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -1145,7 +1129,7 @@

Method Details

}, }, "id": "A String", # Execute the read or SQL query in a previously-started transaction. - "singleUse": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. + "singleUse": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -1176,7 +1160,11 @@

Method Details

"fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. { # Message representing a single field of a struct. "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. + "type": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. # The type of the field. + "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. + "code": "A String", # Required. The TypeCode for this type. + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. + }, }, ], }, @@ -1301,7 +1289,7 @@

Method Details

"resumeToken": "A String", # If this request is resuming a previously interrupted read, `resume_token` should be copied from the last PartialResultSet yielded before the interruption. Doing this enables the new read to resume where the last read left off. The rest of the request parameters must exactly match the request that yielded this token. "table": "A String", # Required. The name of the table in the database to be read. "transaction": { # This message is used to select the transaction in which a Read or ExecuteSql call runs. See TransactionOptions for more information about transactions. # The transaction to use. If none is provided, the default is a temporary read-only transaction with strong concurrency. - "begin": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. + "begin": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Begin a new transaction and execute this read or SQL query in it. The transaction ID of the new transaction is returned in ResultSetMetadata.transaction, which is a Transaction. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -1316,7 +1304,7 @@

Method Details

}, }, "id": "A String", # Execute the read or SQL query in a previously-started transaction. - "singleUse": { # # Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. + "singleUse": { # Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a "negotiation phase" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as "version GC". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table. # Execute the read or SQL query in a temporary transaction. This is the most efficient way to execute a transaction that consists of a single SQL query. "partitionedDml": { # Message type to initiate a Partitioned DML transaction. # Partitioned DML transaction. Authorization to begin a Partitioned DML transaction requires `spanner.databases.beginPartitionedDmlTransaction` permission on the `session` resource. }, "readOnly": { # Message type to initiate a read-only transaction. # Transaction will not write. Authorization to begin a read-only transaction requires `spanner.databases.beginReadOnlyTransaction` permission on the `session` resource. @@ -1348,7 +1336,11 @@

Method Details

"fields": [ # The list of fields that make up this struct. Order is significant, because values of this struct type are represented as lists, where the order of field values matches the order of fields in the StructType. In turn, the order of fields matches the order of columns in a read request, or the order of fields in the `SELECT` clause of a query. { # Message representing a single field of a struct. "name": "A String", # The name of the field. For reads, this is the column name. For SQL queries, it is the column alias (e.g., `"Word"` in the query `"SELECT 'hello' AS Word"`), or the column name (e.g., `"ColName"` in the query `"SELECT ColName FROM Table"`). Some columns might have an empty name (e.g., `"SELECT UPPER(ColName)"`). Note that a query result can contain multiple fields with the same name. - "type": # Object with schema name: Type # The type of the field. + "type": { # `Type` indicates the type of a Cloud Spanner value, as might be stored in a table cell or returned from an SQL query. # The type of the field. + "arrayElementType": # Object with schema name: Type # If code == ARRAY, then `array_element_type` is the type of the array elements. + "code": "A String", # Required. The TypeCode for this type. + "structType": # Object with schema name: StructType # If code == STRUCT, then `struct_type` provides type information for the struct's fields. + }, }, ], }, diff --git a/docs/dyn/spanner_v1.scans.html b/docs/dyn/spanner_v1.scans.html index 61d0572cc50..ac8eee55218 100644 --- a/docs/dyn/spanner_v1.scans.html +++ b/docs/dyn/spanner_v1.scans.html @@ -239,6 +239,7 @@

Method Details

"token": "A String", # The token identifying the message, e.g. 'METRIC_READ_CPU'. This should be unique within the service. }, "startKeyIndex": 42, # The index of the start key in indexed_keys. + "timeOffset": "A String", # The time offset. This is the time since the start of the time interval. "unit": { # A message representing a user-facing string whose value may need to be translated before being displayed. # The unit of the metric. This is an unstructured field and will be mapped as is to the user. "args": { # A map of arguments used when creating the localized message. Keys represent parameter names which may be used by the localized version when substituting dynamic values. "a_key": "A String", diff --git a/googleapiclient/discovery_cache/documents/spanner.v1.json b/googleapiclient/discovery_cache/documents/spanner.v1.json index 8c98cb37195..69a1ea62fa3 100644 --- a/googleapiclient/discovery_cache/documents/spanner.v1.json +++ b/googleapiclient/discovery_cache/documents/spanner.v1.json @@ -2037,7 +2037,7 @@ } } }, - "revision": "20210603", + "revision": "20210624", "rootUrl": "https://spanner.googleapis.com/", "schemas": { "Backup": { @@ -2971,6 +2971,11 @@ "format": "int32", "type": "integer" }, + "timeOffset": { + "description": "The time offset. This is the time since the start of the time interval.", + "format": "google-duration", + "type": "string" + }, "unit": { "$ref": "LocalizedString", "description": "The unit of the metric. This is an unstructured field and will be mapped as is to the user." @@ -4283,7 +4288,7 @@ "type": "object" }, "TransactionOptions": { - "description": "# Transactions Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. # Transaction Modes Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. ## Locking Read-Write Transactions Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. ## Semantics Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. ## Retrying Aborted Transactions When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. ## Idle Transactions A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. ## Snapshot Read-Only Transactions Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. ## Strong Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. ## Exact Staleness These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a \"negotiation phase\" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. ## Bounded Staleness Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. ## Old Read Timestamps and Garbage Collection Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as \"version GC\". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. ## Partitioned DML Transactions Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table.", + "description": "Transactions: Each session can have at most one active transaction at a time (note that standalone reads and queries use a transaction internally and do count towards the one transaction limit). After the active transaction is completed, the session can immediately be re-used for the next transaction. It is not necessary to create a new session for each transaction. Transaction Modes: Cloud Spanner supports three transaction modes: 1. Locking read-write. This type of transaction is the only way to write data into Cloud Spanner. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. 2. Snapshot read-only. This transaction type provides guaranteed consistency across several reads, but does not allow writes. Snapshot read-only transactions can be configured to read at timestamps in the past. Snapshot read-only transactions do not need to be committed. 3. Partitioned DML. This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently. Partitioned DML transactions do not need to be committed. For transactions that only read, snapshot read-only transactions provide simpler semantics and are almost always faster. In particular, read-only transactions do not take locks, so they do not conflict with read-write transactions. As a consequence of not taking locks, they also do not abort, so retry loops are not needed. Transactions may only read/write data in a single database. They may, however, read/write data in different tables within that database. Locking Read-Write Transactions: Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent. Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by Commit or Rollback. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it. Conceptually, a read-write transaction consists of zero or more reads or SQL statements followed by Commit. At any time before Commit, the client can send a Rollback request to abort the transaction. Semantics: Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns `ABORTED`, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner. Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves. Retrying Aborted Transactions: When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous. Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying. Idle Transactions: A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error `ABORTED`. If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., `SELECT 1`) prevents the transaction from becoming idle. Snapshot Read-Only Transactions: Snapshot read-only transactions provides a simpler method than locking read-write transactions for doing several consistent reads. However, this type of transaction does not support writes. Snapshot transactions do not take locks. Instead, they work by choosing a Cloud Spanner timestamp, then executing all reads at that timestamp. Since they do not acquire locks, they do not block concurrent read-write transactions. Unlike locking read-write transactions, snapshot read-only transactions never abort. They can fail if the chosen read timestamp is garbage collected; however, the default garbage collection policy is generous enough that most applications do not need to worry about this in practice. Snapshot read-only transactions do not need to call Commit or Rollback (and in fact are not permitted to do so). To execute a snapshot transaction, the client specifies a timestamp bound, which tells Cloud Spanner how to choose a read timestamp. The types of timestamp bound are: - Strong (the default). - Bounded staleness. - Exact staleness. If the Cloud Spanner database to be read is geographically distributed, stale read-only transactions can execute more quickly than strong or read-write transaction, because they are able to execute far from the leader replica. Each type of timestamp bound is discussed in detail below. Strong: Strong reads are guaranteed to see the effects of all transactions that have committed before the start of the read. Furthermore, all rows yielded by a single read are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Strong reads are not repeatable: two consecutive strong read-only transactions might return inconsistent results if there are concurrent writes. If consistency across reads is required, the reads should be executed within a transaction or at an exact read timestamp. See TransactionOptions.ReadOnly.strong. Exact Staleness: These timestamp bounds execute reads at a user-specified timestamp. Reads at a timestamp are guaranteed to see a consistent prefix of the global transaction history: they observe modifications done by all transactions with a commit timestamp <= the read timestamp, and observe none of the modifications done by transactions with a larger commit timestamp. They will block until all conflicting transactions that may be assigned commit timestamps <= the read timestamp have finished. The timestamp can either be expressed as an absolute Cloud Spanner commit timestamp or a staleness relative to the current time. These modes do not require a \"negotiation phase\" to pick a timestamp. As a result, they execute slightly faster than the equivalent boundedly stale concurrency modes. On the other hand, boundedly stale reads usually return fresher results. See TransactionOptions.ReadOnly.read_timestamp and TransactionOptions.ReadOnly.exact_staleness. Bounded Staleness: Bounded staleness modes allow Cloud Spanner to pick the read timestamp, subject to a user-provided staleness bound. Cloud Spanner chooses the newest timestamp within the staleness bound that allows execution of the reads at the closest available replica without blocking. All rows yielded are consistent with each other -- if any part of the read observes a transaction, all parts of the read see the transaction. Boundedly stale reads are not repeatable: two stale reads, even if they use the same staleness bound, can execute at different timestamps and thus return inconsistent results. Boundedly stale reads execute in two phases: the first phase negotiates a timestamp among all replicas needed to serve the read. In the second phase, reads are executed at the negotiated timestamp. As a result of the two phase execution, bounded staleness reads are usually a little slower than comparable exact staleness reads. However, they are typically able to return fresher results, and are more likely to execute at the closest replica. Because the timestamp negotiation requires up-front knowledge of which rows will be read, it can only be used with single-use read-only transactions. See TransactionOptions.ReadOnly.max_staleness and TransactionOptions.ReadOnly.min_read_timestamp. Old Read Timestamps and Garbage Collection: Cloud Spanner continuously garbage collects deleted and overwritten data in the background to reclaim storage space. This process is known as \"version GC\". By default, version GC reclaims versions after they are one hour old. Because of this, Cloud Spanner cannot perform reads at read timestamps more than one hour in the past. This restriction also applies to in-progress reads and/or SQL queries whose timestamp become too old while executing. Reads and SQL queries with too-old read timestamps fail with the error `FAILED_PRECONDITION`. Partitioned DML Transactions: Partitioned DML transactions are used to execute DML statements with a different execution strategy that provides different, and often better, scalability properties for large, table-wide operations than DML in a ReadWrite transaction. Smaller scoped statements, such as an OLTP workload, should prefer using ReadWrite transactions. Partitioned DML partitions the keyspace and runs the DML statement on each partition in separate, internal transactions. These transactions commit automatically when complete, and run independently from one another. To reduce lock contention, this execution strategy only acquires read locks on rows that match the WHERE clause of the statement. Additionally, the smaller per-partition transactions hold locks for less time. That said, Partitioned DML is not a drop-in replacement for standard DML used in ReadWrite transactions. - The DML statement must be fully-partitionable. Specifically, the statement must be expressible as the union of many statements which each access only a single row of the table. - The statement is not applied atomically to all rows of the table. Rather, the statement is applied atomically to partitions of the table, in independent transactions. Secondary index rows are updated atomically with the base table rows. - Partitioned DML does not guarantee exactly-once execution semantics against a partition. The statement will be applied at least once to each partition. It is strongly recommended that the DML statement should be idempotent to avoid unexpected results. For instance, it is potentially dangerous to run a statement such as `UPDATE table SET column = column + 1` as it could be run multiple times against some rows. - The partitions are committed automatically - there is no support for Commit or Rollback. If the call returns an error, or if the client issuing the ExecuteSql call dies, it is possible that some rows had the statement executed on them successfully. It is also possible that statement was never executed against other rows. - Partitioned DML transactions may only contain the execution of a single DML statement via ExecuteSql or ExecuteStreamingSql. - If any error is encountered during the execution of the partitioned DML operation (for instance, a UNIQUE INDEX violation, division by zero, or a value that cannot be stored due to schema constraints), then the operation is stopped at that point and an error is returned. It is possible that at this point, some partitions have been committed (or even committed multiple times), and other partitions have not been run at all. Given the above, Partitioned DML is good fit for large, database-wide, operations that are idempotent, such as deleting old rows from a very large table.", "id": "TransactionOptions", "properties": { "partitionedDml": {