Decibel! <decibel(at)decibel(dot)org> writes:
> "cache priority" to me sounds like we're trying to influence caching
> behavior, which isn't what's happening. I do agree that we need a
> better way to tell the planner what tables are in memory.
What's been discussed in the past is per-tablespace settings for
random_page_cost and friends. That was meant to cover actual disk
hardware differences, but could be (ab)used to handle the case of
heavily and not so heavily used tables.
Per-table sounds kinda bogus to me; such settings would probably reflect
wishful thinking on the part of the DBA more than reality.
regards, tom lane