Updated on 2026-01-06 GMT+08:00

Data Sharing Functions

pgxc_group_add_subscription(src_vw_name, target_vw_name)

Description: In a storage-compute decoupling design, logical clusters (Virtual Warehouses or VWs) create a KV subscription link. After setup, the consumer VW's KV cache regularly updates by syncing incremental SST files from the producer's OBS cudesc snapshot folder to its local system. This enables cross-VW queries on tables stored in the producer VW. The created subscription relationship can be found in the pgxc_group_subscription table. This function is supported only by clusters of version 9.0.3 or later.

Return type: void

The following information is displayed:

Table 1 pgxc_group_add_subscription(src_vw_name, target_vw_name) return columns

Column

Type

Description

src_vw_name

text

Producer VW name, which is usually used as the VW to which data is written.

target_vw_name

text

Consumer VW name, which is usually used as the VW for reading data.

The following is an example:

1
2
3
4
5
SELECT pgxc_group_add_subscription('write_group', 'single_read_group');
pgxc_group_add_subscription 
------------------

(1 row)

pgxc_group_drop_subscription(src_vw_name, target_vw_name)

Description: To delete an established KV subscription relationship in a storage and compute separation architecture, remove the records from the pgxc_group_subscription table. After deletion, the consumer VW can no longer query tables in the producer VW across VWs. This function is supported only by clusters of version 9.0.3 or later.

Return type: void

The following information is displayed:

Table 2 pgxc_group_drop_subscription(src_vw_name, target_vw_name) return columns

Column

Type

Description

src_vw_name

text

Producer VW name, which is usually used as the VW to which data is written.

target_vw_name

text

Consumer VW name, which is usually used as the VW for reading data.

The following is an example:

SELECT  pgxc_group_drop_subscription('write_group', 'single_read_group');
pgxc_group_drop_subscription 
------------------------------

(1 row)