Re: largest deployment?

Some of my preliminary tests with squid today proved very interesting.
I have a PPro 180 (440FX) 64MB EDO RAM, 4GB Ultra DMA drive, noatime,
Linux 2.0.33, 32MB cache, Squid 1.1.20, hacked to make a "comm*"
connection only to the accel server, disabled DNS. I was sustaining
250 hits per second on cached data. This is configured as an httpd
accellerator.

The client machine was a identical machine running 504 custom clients
that retrieve a list of URLs in sequential fashion as fast as possible.
They were connected via a 100BaseTX crossover cable with 3c905
Vortex/Boomerang NICs.

I am estimating that the machine's capacity is 21.5 million hits per
day maximum, and 11.5 million minimum. I plan on utilizing 10 to
begin with, with an estimated peak capacity of 215 million hits per
day.

I'm still waiting to try 1.2 and Linux 2.2 on these machines...with all
the new hashes and stuff, 2.2 should REALLY scream!

--Perry

>
> Dancer <dancer@brisnet.org.au> writes:
>
> > 34GB/server, split across half a dozen drives, sharing by squid (no
> > raid). No problems with the ext2 file-system so far. Bearing up nicely.
> >
> > D
>
> Ah, just edge you out. :) 38 gig, P-II 266, 512meg ram, running
> squid1.2. Handing peak loads of 150 tcp/second, 60 min averages in the
> 50 tcp/second.
>
> Michael.
>
>
>

-- 
Perry Harrington        System Software Engineer    zelur xuniL  ()
http://www.webcom.com  perry.harrington@webcom.com  Think Blue.  /\