Re: Sun vs. Mac - best Postgres platform? - Mailing list pgsql-performance

From Andrew Sullivan
Subject Re: Sun vs. Mac - best Postgres platform?
Date
Msg-id 20030114141853.V5335@mail.libertyrms.com
Whole thread Raw
In response to Re: Sun vs. Mac - best Postgres platform?  (CaptainX0r <captainx0r@yahoo.com>)
List pgsql-performance
On Tue, Jan 14, 2003 at 11:01:49AM -0800, CaptainX0r wrote:

> I changed back to the default 1024, and down to the minimum, 64
> - no change.  I think that was changed simultaneously with some
> other parameter (bad, I know) that actually had an affect.  I
> guess I can remove it.

Very bad to change two things at once.  You think it's saving you
time, but now . . . well, you already know what happens ;-)  Anyway,
you _still_ shouldn't have it that high.

> > > effective_cache_size = 65536    # typically 8KB each
>
> I read somewhere that this should be set to half the system RAM
> size, 64k*8k=512m = 1/2 of the 1 Gig RAM.  I guess this is way
> off since you're saying that it's disk cache.  This agrees with
> the documentation.  I can't really rely on the (precious little
> Solaris postgres) info I find on the net....  ;)

I think you should rely on the Postgres documentation, which has way
fewer errors than just about any other technical documentation I've
ever seen.  Yes, it's disk cache.

I wouldn't set _anything_ to half the system RAM.  It'd be real nice
if your disk cache was half your RAM, but I'd be amazed if anyone's
system were that efficient.

It sounds like you need to follow Tom Lane's advice, though, and do
some profiling.

A
--
----
Andrew Sullivan                         204-4141 Yonge Street
Liberty RMS                           Toronto, Ontario Canada
<andrew@libertyrms.info>                              M2P 2A8
                                         +1 416 646 3304 x110


pgsql-performance by date:

Previous
From: CaptainX0r
Date:
Subject: Re: Sun vs. Mac - best Postgres platform?
Next
From: CaptainX0r
Date:
Subject: Re: Sun vs. Mac