David Jaquay <djaquay(at)gmail(dot)com> writes:
> Summary: Doing a two or three table join for a date range performs
> worse than doing the same query individually for each date in the
> range.
Could we see EXPLAIN ANALYZE, not just EXPLAIN, results?
Also, have you ANALYZEd lately? If the estimated row counts are at all
accurate, I doubt that forcing a nestloop indexscan would improve the
situation.
Also, what PG version is this?
regards, tom lane