Lance,
> I have a PostgreSQL database that runs on a dedicated server. The
> server has 24Gig of memory. What would be the max size I would ever
> want to set the shared_buffers to if I where to relying on the OS for
> disk caching approach? It seems that no matter how big your dedicated
> server is there would be a top limit to the size of shared_buffers.
There's not, actually. Under some circumstances (mainly Solaris 10 + UFS
on AMD) it can actually be beneficial to have s_b be 80% of RAM and bypass
the FS cache entirely. This isn't usually the case, but it's not to be
ruled out.
If you're relying on the FS cache and not using direct I/O, though, you
want to keep at least 50% of memory free for use by the cache. At below
50%, you lose a significant part of the benefit of the cache without
losing the cost of it. Of course, that assumes that your database is
bigger than ram; there isn't much need to have either s_b or the f.s.c. be
more than twice the size of your whole database.
In general, a setting s_b to 25% of RAM on a dedicated machine, and 10%
(with a max of 512MB) on a shared machine, is a nice safe default which
will do OK for most applications.
--
--Josh
Josh Berkus
PostgreSQL @ Sun
San Francisco
