[CapriciousCptl]

You can fiddle with the autovacuum daemon[1,2] but we've never really had to. These days we just run AWS RDS when it counts or a dedicated VPS when it doesn't and things go fine--

[1,2] https://www.postgresql.org/docs/13/routine-vacuuming.html https://www.postgresql.org/docs/current/planner-stats.html

The main issue we get is the 1 connection = 1 process issue although there are ways to mitigate that (namely pgbouncer).