Re: Performance Tuning

From: Ilja Pavkovic <illsen@dont-contact.us>
Date: Thu, 25 May 2000 21:56:43 +0200

----- Original Message -----
From: Erik Horn <Erik_Horn@beavton.k12.or.us>
To: <squid-users@ircache.net>
Sent: Thursday, May 25, 2000 6:41 PM
Subject: Performance Tuning

> I have a couple of performance tuning questions, and would like to get any
> opinions that would help me to make this cache as fast and efficient as
> possible.
>
> I am running Redhat 6.1 on a Dell Poweredge 4400 server with a 667Mhz P3X,
> 512MB (133MHz SDRAM), and 5 x 9gb 10k rpm U160 drives connected to a raid
> controller with 64 meg ram and configured for write caching. 1 GB of each
> drive is striped with raid 5 for the operating system. The rest of each
> drive is set up as a standalone volume to create an OS volume of 4GB and 5
> x 8GB cache volumes.
>
> I am running squid 2.2s5-hno with delay-pools, memory pools, async-io (32
> threads), and dl-malloc. There are 5 cache_dirs, each with 6200kb, 16/250
> directories. Cache_mem is 64MB. During business hours, the cache is
> averaging about 700 concurrent open files (configured for 16000 maximum).
> During peak times I've seen the request rate go up to 65/sec (6900/min).

- mount all the discs containing your cache_dirs with the option "noatime"
- renice your squid process to -20.
- install cachemgr.cgi to squid statistics.

Ilja Pavkovic
Received on Thu May 25 2000 - 14:15:19 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:53:34 MST