[SDK]: Client support partition key

[xige-16]

Is there an existing issue for this?

• I have searched the existing issues

Is your feature request related to a problem? Please describe.

I would like to introduce another level of abstract, partition key.
So when user create a collection

from pymilvus import CollectionSchema, FieldSchema, DataType
book_id = FieldSchema(
name="book_id",
dtype=DataType.INT64,
is_primary=True,
)
book_name = FieldSchema(
name="book_name",
dtype=DataType.VARCHAR,
max_length=200,
is_partition_key=true,
)
book_intro = FieldSchema(
name="book_intro",
dtype=DataType.FLOAT_VECTOR,
dim=2
)
default_schema = CollectionSchema(
fields=[book_id, book_name, book_group, book_intro],
description="Test book search",
)

collection = Collection(name="hello_milvus", 
schema=default_schema, 
shards_num=1, 
num_partitions=1024)

So you will see book group is the parititon key, and the collection will create 1024 partitions by default.

When insert happens:

hash(book_name)%1024 to decide which segment should be written into
others are exactly same as current insert path.

When search:

User must specify a group of partition key when search (Performance may degrade if not specified)
Milvus first pick the correct segment to work with
Then do scalar filtering on the segment based on partition keys and other filter conditions.

Describe the solution you'd like.

@XuanYang-cn @longjiquan @congqixia @yhmo @yelusion2 @shanghaikid @yah01 @nameczz

• [ pymilvus ] [FEATURE]: Support logical partition keys pymilvus#1458

Describe an alternate solution.

No response

Anything else? (Additional Context)

No response

[yhmo]

milvus-io/milvus-sdk-java#492
Already done./