Tables distribution questions

[yinan8128]

Hi team,

After checking 12.1 doc, we still have questions about tables distribution. We raised questions in Slack but didn't get needed answer

Please help provide answer, thanks:

For example, we have tables with tenant_uuid column, if we distribute tables as below to make them work co-location

SELECT create_distributed_table('table_1', 'tenant_uuid');
SELECT create_distributed_table('table_2', 'tenant_uuid', colocate_with => 'table_1');
SELECT create_distributed_table('table_3', 'tenant_uuid', colocate_with => 'table_1');
...
load data or logical replication from source database

1. Does table_1 need to be the first table which have new tenant_id record(earlier than other tables table_2, table_3...)?
2. Does table_1 need to have full tenant_id record(not allow table_2, table_3... have the tenant_id which not in table_1 table)?

Thank you

[emelsimsek]

• Does table_1 need to be the first table which have new tenant_id record(earlier than other tables table_2, table_3...)?

No, not necessarily for the distribution/colocation purposes. But if your data model requires that, you may make tenant_uuid a primary key for table_1.

• Does table_1 need to have full tenant_id record(not allow table_2, table_3... have the tenant_id which not in table_1 table)?

Again, if your data model requires this relationship, you may define a foreign_key constraint for tenant_uuid in table_2 and table_3 that references tenant_uuid in table_1.

You may colocate the tables without those constraints as well.