Message-Id: <4728B341.EE98.0025.0@wicourts.gov>
Date: Wed, 31 Oct 2007 16:54:26 -0500
From: "Kevin Grittner" <Kevin.Grittner@wicourts.gov>
To: "Ketema" <ketema@gmail.com>,<pgsql-performance@postgresql.org>
Subject: Re: Hardware for PostgreSQL
References: <1193849108.591961.103570@22g2000hsm.googlegroups.com>
In-Reply-To: <1193849108.591961.103570@22g2000hsm.googlegroups.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

>>> On Wed, Oct 31, 2007 at 11:45 AM, in message
<1193849108.591961.103570@22g2000hsm.googlegroups.com>, Ketema
<ketema@gmail.com> wrote:=20
=20
> Who has built the biggest baddest Pg server out there and what do you
> use?
=20
I don't think that would be us, but I can give you an example of
what can work.  We have a 220 GB database which is a replication
target for OLTP and supports a web site with about 2 million web
hits per day.  Daily, we have about 1 million database transactions
which modify data and 10 million database transactions which are
read-only.  The box has 8 4 GHz Xeon processors, 12 GB RAM, and RAID
5 with 13 live spindles and two hot spares.  The RAID controller has
256 MB RAM with battery backup.  This gives very good performance,
properly tuned.
=20
Besides the PostgreSQL database, the box also runs middle tier
software written in Java.
=20
I'll second the opinions that connection pooling and a high-
quality RAID controller with battery backed RAM cache are crucial.
=20
-Kevin
=20