Re: [squid-users] Request Per Second

On Sun, May 09, 2004 at 02:52:48PM -0500, Matt wrote:
> When using a fast 7200 RPM 8.9ms seek IDE drive with Squid does this
> effectively create a limit to the number of requests per second the cache is
> able to support? Since 8.9ms x 100 = 890ms does this effectively limit you
> to 100 requests per second? When you begin to hit that limit will CPU load
> begin to spike as a sign of it?

I guess that this calculation is a little too easy. :) We ran 100 req/sec
on a Pentium III/800 MHz with a RAID-1 cluster of old hard disks and hat
a load of 0.3 on the system. 8.9ms is just the time your hard disk needs
to read a single block. Squid keeps some management information in memory. And
loading larger objects will take longer than that. Then again only very few
objects are cached since web server administrators hardly care about web
caches... many factors... just try it.

 Christoph

-- 
~
~
".signature" [Modified] 3 lines --100%--                3,41         All