From: Alexander Priem <ap@cict.nl>
To: pgsql-performance@postgresql.org
Subject: Tuning PostgreSQL
Date: Mon, 21 Jul 2003 10:31:19 +0200
Message-ID: <002a01c34f62$75c01de0$b696a8c0@APR> (raw)
Hi guys,
I am new to PostgreSQL and have done some "extensive" research already. If you could give me some advice/confirmation, I would be really grateful.
I am going to build a PostgreSQL database server for a client. This database will contain many tables (over 100, maybe more), with some of them containing over 1 million records pretty soon. This database will be accessed via a desktop application (Windows) and a web application (PHP/Apache). There may be over 100 people accessing the database at any given time (selecting, inserting and updating), but I don't think it will be a lot more than 100 at any given time. Most of the time, it will be less.
What I am thinking about is buying a server with the following specifications:
* 1 or 2 Intel Xeon processors (2.4 GHz).
* 2 Gigabytes of RAM (DDR/ECC).
* Three 36Gb SCSI160 disks (10.000rpm) in a RAID-5 config, giving 72Gb storage space (right?). The RAID-5 controller has a (hardware) cache of 128Mb.
* 100Mbit ethernet.
I will run RedHat Linux 9 (kernel 2.40) with PostgreSQL 7.3.3 on this server.
What would you think of this hardware config? Would it do? Of would 4Gb RAM be a lot better? What do you think about the need for two Xeon procs?
Finally, I have some questions about postgresql.conf (who doesnt?). After some research, I think I will go for the following settings initially. Maybe benchmarking will lead to other values, but I think these settings will be a fine starting point :
shared_buffers = 6000 (kernel.shmmax = 60000000)
sort_mem = 4096
max_connections = 150
vacuum_mem = 65536
What do you think of these settings? Do you have any other hints for optimizing PostgreSQL
Many many thanks in advance :)
Kind regards,
Alexander Priem
CICT Solutions
Email: ap@cict.nl
Internet: www.cict.nl
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: ap@cict.nl
Subject: Re: Tuning PostgreSQL
In-Reply-To: <002a01c34f62$75c01de0$b696a8c0@APR>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox