[squid-users] squid 2.5 rproxy performance - seems good.. tips for web-polygraph?

Hi guys,

I've set up a squid 2.5 with the rproxy.patch.. as a reverse proxy for our
sites..

The HW setup is an IBM eServer x330 - 2x 1.26ghz PIII, 512kb L2 (actually
the one tested below was 2x1ghz, 256kb l2), 1gb ram and 2x 18gb disks in HW
Raid 1.

when I "accidently" had to do a realistic test - squid served approx.
900/sec.
The reason seemed to be the amount of interrupts.. it was maxing out at
18,300 interrupts/sec.

My guess is that this was due to me using the onboard netcard.

I'm going to try to get hold of the Network card IBM uses for all the
SpecWeb tests.. an AceNic adapter.

I've tried to use ab -n 10000 -c 1000 up against the squid (1ghz model)
with an object size of 500 bytes - so as not run into network constraints.

The squid served 2200 req/sec. - I compared this with the same page gotten
from a CacheFlow 725 box and it served 3100 req/sec.

I hope, that when I insert the AceNic adapter to my squid box, It will
deliver approx. 2000 req/sec - only dependant on the cpu-load.

I'm gonna run a web-polygraph test against the squid and the cacheflow - to
compare realistic price/performance.

Do any of you have any tips for a good web-polygraph config I could/should
use?

Thanks in advance.

-------------| This mail has been sent to you by: |------------
Klavs Klavsen, IT-coordinator and Systems Administrator at
Metropol Online - http://www.metropol.dk
Tlf. 33752700, Fax 33752720, Email ktk@metropol.dk

Private- Email [email protected] - http://www.vsen.dk

--------------------[ I believe that... ]-----------------------
It is a myth that people resist change. People resist what other
people make them do, not what they themselves choose to do...
That's why companies that innovate successfully year after year
seek their peopl's ideas, let them initiate new projects and
encourage more experiments. -- Rosabeth Moss Kanter
Received on Fri Nov 23 2001 - 01:45:23 MST