Anyone happen to know if for a partitioned table in #PostgreSQL (BY LIST (dataset_id)) does each partition have its own separate index, even if that index comes from an extension (#PGVectorScale)? Meaning instead of having a table per dataset, if I use one table, but with partitions, is only the index of the one partition scanned when WHERE dataset_id = $1? I suspect partitions also count against the limit of relations? I tried to ask the creators of the extension, but got no reply. #postgres

To #PostgreSQL experts: Is there a downside on having *a lot* of tables (hundreds, maybe thousands in future) where they all have foreign keys with ON DELETE CASCADE to 4 central tables? The tables are #PGVector (#PGVectorScale) datasets, separated because #SQL WHERE is applied by PGVectorScale only after the vector search. Also the different datasets might have different vector dimensions in the future. (Also never did any clustering and wonder if its good or bad for that.)