"Josh Berkus" <josh(at)agliodbs(dot)com> writes:
> What have other Postgres users done to speed up aggregates on large
> tables?
FWIW, I've implemented hashed aggregation in CVS tip. I have not had
the time to try to benchmark it, but I'd be interested if anyone can
run some tests on 7.4devel. Eliminating the need for a SORT step
should help aggregations over large datasets.
Note that even though there's no SORT, the sort_mem setting is used
to determine the allowable hashtable size, so a too-small sort_mem
might discourage the planner from selecting hashed aggregation.
Use EXPLAIN to see which query plan gets chosen.
regards, tom lane