Feature: Table Cluster Key

[BohuTANG]

Summary

Goals for reduce the partition scans.
In Fuse engine, physical data will be stored follow the append order, if one column data is scattered across multiple partitions, the query related to this column will be bad.

ALTER TABLE t1 CLUSTER BY (column1, column2)
When the table is re-clustered, new partitions will be created and the physical data will re-partition by column1 and column2.

This is same as snowflake did.

Reference:
https://docs.snowflake.com/en/user-guide/tables-clustering-keys.html
https://docs.snowflake.com/en/user-guide/tables-clustering-micropartitions.html#label-data-clustering
https://docs.snowflake.com/en/sql-reference/functions/system_clustering_information.html

[zhyass]

#4946