Hello Hannu,
Friday, July 09, 1999 you wrote:
H> Still I'm quite sure that the main effort in PostgreSQL development has
H> so
H> far gone to optimising queries where most of the data is fetched from
H> the
H> disk.
Oh, I see. This is appropriate for some not time critical
and dumb applications, such as web DB. But this is out of the
way of speed server tasks. Maybe Postgres has been designed with
such plan in mind - to use big DBs from disc? That is not
good news for me either. Almost everyone has suggested me
to use more RAM to speed up queries, and now it turned out to
be not in Postgres's mainstream. Maybe there is something
wrong with this ideology, since RAM is bigger and cheaper
every day?
H> forward-id-links on frequently updated DBs. On infrequently updated DBs
H> you could just use triggers and/or cron jobs to keep your reports
H> updated,
H> I quess that this is what most commercial OLAP systems do.
It seems that trigger will be the last resort.
Best regards, Leon