From: | "Richard Huxton" <dev(at)archonet(dot)com> |
---|---|
To: | "Dr R(dot)Adscheid" <adscheid(at)rosin(dot)com> |
Cc: | <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: vacuum |
Date: | 2001-01-26 09:35:44 |
Message-ID: | 009201c0877b$7f9b8680$1001a8c0@archonet.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
"Dr R.Adscheid" <adscheid(at)rosin(dot)com> wrote in message
news:94rb3d$f5v$1(at)news(dot)tht(dot)net(dot)(dot)(dot)
> We are using PostgreSQL 6.3/Digital Unix 4.0B in an environment with 7x24
> availiability. There
You might want to consider upgrading when possible - I think there have been
fairly substantial changes since 6.3
> is one table, which has about 9000 new records per day and about 10% being
> updated. With an index over several columns the select on this table is
quit
> short, but removing old entries and vacuuming is an very time
> consuming operation (about 1 hour for the whole database!) and because of
One hour to vacuum 9000 records seems to be a *very* long time. Almost e
faster to do it by hand. You aren't short of RAM? Actually - you say that's
the whole database, so it might be reasonable - depends on what's in the
rest.
> the 7x24 production not acceptable. On the other hand, no index improves
the
> removing and vacuuming, but now the select is very time consuming, which
is
> also not acceptable. Even with the best solution (some index, which
improves
> the select but slows down the cleaning) our customer complaints .
> The best way to solve this, would be to remove the feature of keeping
> deleted/updated records in the databasefiles and therefor no need to
vacuum.
> Is there any way to configure this when compiling? Or are there other
> possibilities?
Try dropping the index, vaccuming, recreate the index. Might well be a lot
quicker.
You might also find an index on 2 or 3 columns gives you selects that are
almost as fast, but speeds inserts/updates.
- Richard Huxton
From | Date | Subject | |
---|---|---|---|
Next Message | Gilles DAROLD | 2001-01-26 10:22:13 | Re: Connection pooling |
Previous Message | Igor V. Rafienko | 2001-01-26 09:33:19 | Re: The type int8 and the use of indexes |