zlacker

> "But making sure that PostgreSQL’s autovacuum is running as best as possible is difficult due to its complexity."

The problem, as the article states it, is that a "sensible" vacuum setting for one table is a terrible setting for another depending on how large these tables are. On a 100 million tuple table you'd be waiting 'til there there were 20 million garbage tuples before taking action.

>>spudly+(OP)
You can set autovacuum reloptions on a per-table basis, if they differ that much for your your use case.