Pawel Veselov <pawel(dot)veselov(at)gmail(dot)com> writes:
> I was wondering how come there is such a drastic difference between finding
> max and min. Seems like "index scan backwards" is really bad...
It's probably an artifact of your data distribution, ie, the "blockid =
4814" condition is skipping lots of rows at one end of the index but few
or none at the other.
If you're concerned about the performance of this type of query, an index
on (blockid, rowdate) would work a lot better than the ones you've
provided.
regards, tom lane