"Robert Haas" <robertmhaas@gmail.com> writes:
> It seems very strange for the planner to decide to build an in-memory
> hash table on a column that is already indexed (the primary key, no
> less!).
What's strange about it? A probe into an in-memory hashtable is a lot
cheaper than a probe into an index, so this type of plan makes plenty
of sense if the hashtable will fit in RAM and there are going to be a
lot of probes. (Where "a lot" means "enough to amortize the cost of
building the hashtable", of course.)
> Experimentation shows this is actually about 25% faster.
Well, that just says your cost parameters need a bit of adjustment
if you'd like the planner to get the crossover point exactly right.
regards, tom lane
