Jean-Michel Pouré<jm(at)poure(dot)com> wrote:
> [no postgresql.conf changes except]
> shared_buffer 24M.
That's part of your problem. (Well, that's understating it; we don't
have any real evidence that you have any performance problems *not*
resulting from failure to do normal configuration.) If you download
the configurator tool I referenced in another email you can get a
start on fixing this. Or search the lists -- this stuff had been
discussed many, many times. Or do a google search; I typed
postgresql performance configuration
and got many hits. If you spend even one hour closely reviewing any
of the top few hits and testing a more optimal configuration for your
hardware and workload, most of your performance problems would
probably go away.
Then, if you *still* have any performance problems, people could help
you diagnose and fine-tune from a base which should be much closer.
The truth is, with a proper configuration your "biggest problem",
which you've misdiagnosed as the result of casting, would probably go
away. I will take another look at it now that you have the results of
EXPLAIN ANALYZE posted, but I seriously doubt that it's going to do
well without tuning the configuration.
pgsql-performance by date
|Next:||From: Kevin Grittner||Date: 2009-08-26 23:27:29|
|Subject: Re: PostgreSQL does CAST implicitely between int and a domain derived from int|
|Previous:||From: Guillaume Smet||Date: 2009-08-26 21:59:25|
|Subject: Re: Performance regression between 8.3 and 8.4 on heavy text indexing|