Kevin Grittner wrote:
> Claire Chang <yenhsiac@yahoo.com> wrote:
>
>> shared_buffers = 32GB
>>
>
> I seem to remember seeing some benchmarks showing that performance
> falls off after 10GB or 20GB on that setting.
>
Not even quite that high. I've never heard of a setting over 10GB being
anything other than worse than a smaller setting, and that was on
Solaris. At this point I never consider a value over 8GB, and even that
needs to be carefully matched against how heavy the writes on the server
are. You just can't set shared_buffers to a huge value in PostgreSQL
yet, and "huge" means ">8GB" right now.
Note that the problems you can run into with too much buffer cache are
much worse with a low setting for checkpoint_segments...and this
configuration doesn't change it at all from the tiny default. That
should go to at least 64 on a server this size.
--
Greg Smith 2ndQuadrant US greg@2ndQuadrant.com Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support www.2ndQuadrant.us