On 6/17/05, Pieterjan Heyse <pieterjan.heyse@xxxxxxxxxxxxxxxxxxxxx> wrote:
> We are a school with 1200 students, and 200 pc's running squid +
> Dansguardian as proxying service. The http requests tend to be slow at
> peak moments and I am wondering if I could resolv this by changing the
> setup.
I would start with some work to investigate the root cause of the slow
response at peak times. You might begin by monitoring both your
Internet connection utilization, the network port(s) the Squid server is
connected to, and disk+CPU performance in the OS on the Squid host.
All of the above is out of the scope of the Squid mailing list. You might
consult a support list for the OS you use on Squid, and look into software
to collect statistics from your switch and router (MRTG, Cricket, etc).
> Currently we have a celeron 900 pc with 1 ATA disk acting as a proxy.
> A nearby factory donated an old dual P2-300 server with 7 SCSI disks
> 4GB each. I read that performance of squid increases with every disk
> you add, so will the switch to this slower (cpu wise) system be
> beneficial to us, ro should I stick to the celeron 900 and add some
> more RAM ?
>
> The celeron has 256MB RAM, the donated server has 512MB RAM.
Assuming that Squid+Dansguardian is not bound by CPU speed (you
can watch CPU utilization on the base OS using top, iostat, etc) then
moving to more/faster disks will improve serving results out of cache,
and more memory is always beneficial.
> Are there ways to benchmark squid in an easy way ?
There are a number of scripts to parse Squid's access_log and report on
usage and performance:
http://www.squid-cache.org/Scripts/
There are also tools you can use to load-test any HTTP or proxy server,
artificially generating many thousands of requests and tracking response time.
Kevin Kadow
