[squid-users] Help to get the maximum squid setup

Hi,
I have been testing squid on a test machine and is now ready to be used in an
IBM xseries 205 with 1gig of RAM, 2.4 ghz CPU and two(2) 18.2 10k rmp SCSI
disks.

Previously, I have used this configuration:

./configure --prefix=/usr/local/squid \
--enable-async-io \
--with-pthreads \
--enable-storeio="aufs,ufs"
--enable-snmp \
--disable-http-violations \
--disable-hostname-checks \
--enable-underscores \
--enable-auth="basic" \
--enable-basic-auth-helpers="NCSA"
--enable-default-err-language=English

Operating System would be Slackware Linux, kernel 2.4.22. Should I stay with
pthreads or should I use the included dlmalloc in squid? Which is better in
performance?

Also what filesystem would you guys recommend? I am planning to use XFS
although the test machine is now running on ReiserFS, does XFS has a
performance advantage over ReiserFS according to your experience?

Any other tips or suggestions or anything that I have to consider to achieve
maximum performance on the machine? Estimated number of users is 100 to 200
only.

Thanks and best regards,
Kenneth Oncinian
Received on Tue Sep 30 2003 - 03:04:04 MDT