From: | "Ross J(dot) Reedstrom" <reedstrm(at)rice(dot)edu> |
---|---|
To: | Jeff Boes <jboes(at)nexcerpt(dot)com> |
Cc: | pgsql-admin(at)postgresql(dot)org |
Subject: | Re: Poor performance of UPDATE against busy table |
Date: | 2001-09-06 19:15:22 |
Message-ID: | 20010906141522.A3371@rice.edu |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
You don't mention your version of pgsql, but if you get a lot of
throughput on that table, how often do you VACUUM it? All those dead
tuples take space that has to be scanned through, since indices
don't keep track of tuple state.
Ross
On Thu, Sep 06, 2001 at 02:31:51PM -0400, Jeff Boes wrote:
> We have a table being used as a job queue. There's one 'scheduler'
> process that manipulates jobs (inserting, updating, and deleting rows/jobs as
> they change status). There are a number (~24) of processes reading and
> updating rows as they 'claim' and 'complete' jobs.
>
> Our problem is that throughput on this table is fairly low. Updates by
> the 'scheduler' sometimes take 15 seconds (!!). We've tried various
> combinations of LOCK TABLE commands, but without much success.
>
> Does anyone know if there are ways to improve the performance of UPDATE
> against a table where there are many readers?
>
>
> --
> Jeff Boes vox 616.226.9550
> Database Engineer fax 616.349.9076
> Nexcerpt, Inc. jboes(at)nexcerpt(dot)com
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: Have you searched our list archives?
>
> http://www.postgresql.org/search.mpl
From | Date | Subject | |
---|---|---|---|
Next Message | Jeff Boes | 2001-09-06 19:44:48 | Re: Poor performance of UPDATE against busy table |
Previous Message | Jeff Boes | 2001-09-06 18:31:51 | Poor performance of UPDATE against busy table |