Joseph Shraibman <jks(at)selectacast(dot)net> writes:
> How do I read that? Is it creating a hash out of the data in d, then
> going through u doing a join?
Yeah. Given the numbers of rows involved, the plan seems pretty
reasonable --- I doubt you can do a lot better within the context
you're showing here. To make it faster you'll have to find a way
to not need to look at all the rows.
regards, tom lane