From: | Greg Smith <greg(at)2ndquadrant(dot)com> |
---|---|
To: | Dave Page <dpage(at)pgadmin(dot)org> |
Cc: | "Massa, Harald Armin" <chef(at)ghum(dot)de>, "A(dot)Bhattacharya" <A(dot)Bhattacharya(at)sungard(dot)com>, pgsql-general <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: Postgres.exe Process taking too much memory and CPU usage - making the system extremely slow. |
Date: | 2009-12-11 15:11:01 |
Message-ID: | 4B226105.4040208@2ndquadrant.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Dave Page wrote:
> On Fri, Dec 11, 2009 at 9:02 AM, Massa, Harald Armin <chef(at)ghum(dot)de> wrote:
>
>> I got MUCH better results by drastically lowering shared_buffers on
>> Windows. Drastically = 8MB.
>>
>
> Wow - really? Greg and I did some rough pgbench experiments last year
> and were finding the on a 4GB machine, running XP Pro, 512MB seemed to
> be optimal, but it was only marginally better than 128 or 256MB. Going
> lower than that made a noticable difference, and going higher we saw
> performance dropping off again as well.
>
Dave is talking about Greg Stark here, lest anyone credit me for
something I wasn't involved in. I just updated the shared_buffers
section of "Tuning Your PostgreSQL Server" to reflect Dave's comments
and to generally clean up the Windows suggestions here. I recall seeing
some comments in the past that suggested earlier systems started to fall
off at closer to 64MB rather than 128MB, tweaked the wording there
accordingly. 8MB working out best is really unexpected though; I'd like
to know what you were doing where *that* was the optimal setting.
--
Greg Smith 2ndQuadrant Baltimore, MD
PostgreSQL Training, Services and Support
greg(at)2ndQuadrant(dot)com www.2ndQuadrant.com
From | Date | Subject | |
---|---|---|---|
Next Message | Adrian Klaver | 2009-12-11 15:11:06 | Re: pg_dump and ON DELETE CASCADE problem |
Previous Message | Robert Gottstein | 2009-12-11 15:09:26 | Transaction read command fetching of single Data Items |