Sean Davis <sdavis2@mail.nih.gov> writes:
> You might think about full-text searching of the descriptions. It may be
> quite a bit faster than a "like" search. It may be worth looking.
It's not the LIKE search that's her problem though --- the time is going
into fetching the matching rows from the other table.
If the "go" table becomes large then it'd be time to worry about that,
but right now it's only costing her 80 msec to seqscan it...
regards, tom lane