Re: [Fwd: Optimizing Squid performance]

If you're using 2.4, a good place to start is
on IPC tunning. There's a good description under
"diskd" on the FAQ.
        If your want to put your mind to rest,
you could just run a pair o polygraph testers to see
where's the breakdown point (with a RAID sub-system
and one CPU, probably well bellow 150req/sec...

--jp

Athanasios Kouroussis wrote:
>
> Thanks for that.
>
> Returning to my original question, which parameters should I tweak in order
> to achieve max. performance (this is an item nobody responded, except from
> the RAID 5 advice and explanation by Robert Collins).
>
> I repeat my the configuration of my box: Compaq Proliant 1600, 1 PIII
> 600Mhz, 512Mb RAM, 4x9Gb SCSI-3 disks in RAID 5 (soon to be 6 disks in RAID
> 0+1) running Linux 2.2.14. Squid 2.4devel3-ntlm (1 20Gb cache dir). I use
> SquidGuard (10 children) for filtering and a home made authenticator (10
> children) for authorizing valid NTLM users.
>
> Thanks in advance,
> Athanasios Kouroussis
> Unifon - T.C.P. S.A.
> akouroussis@unifon.com.ar
>
> Henrik Nordstrom wrote:
>
> > Athanasios Kouroussis wrote:
> >
> > > 1) Top or free report 3 of 512Mb of physical memory free. Only
> > > 40Mb of swap space used.
> >
> > Quite normal regardless of the amount of memory you have as all the free
> > memory will be used ass buffer/cache memory to speed up disk accesses.
> >
> > See "free" for a more detailed output.
> >
> > Squid will be quite happy as long as it fits completely in memory, and
> > the machine is not swapping. So your machine should be equipedd with
> > little more memory than the size of the Squid process. How big margin is
> > required depends on your traffic rate.
> >
> > --
> > Henrik Nordstrom
> > Squid hacker

--
FCCN - Fundação para a Computação Científica Nacional - Tel:
+351-218440100