On 14 Nov, Henrik Steffen wrote:
> of course, there are some seq scans... one of the most
> difficult queries is for example a kind of full text
> search, that searches through 8 different tables with
> each between 300.000 and 500.000 rows and 5-50 columns,
> but that's a different issue (need a full-text-search-engine...)
Ah, well, it may be worthwhile to check out fulltextindex or tsearch
in contrib/. They both require some changes to the way you do queries,
but they may be helpful in speeding up those queries.
-johnnnnnnnnn