I'm looking for a proven set of "configure" options to build SQUID 2.5-Stable 10 and a SQUID.CONF file that will give the best performance while conserving the bandwidth on our T-1's. Our network topology is 16 spokes connected to a hub site by Point-To-Point T1's. The network is home to 2,000 Windows desktops, 700 Windows laptops and 50 Windows servers. System wide, the web traffic is a 40/60 ratio static and dynamic pages. We have 2 Squid servers in a round-robin configuration at the hub site connecting to our ISP and 12 of the spoke sites have a "child" Squid running. At the hub site our server hardware is dual 800MHz P3 CPU's, 1GB RAM, 40GB IDE disk, 100MB/s switched Ethernet. The software platform is SuSE 9.2 Pro (no GUI), Squid 2.5 STABLE 9 and SuSE Firewall. At 12 of the spoke sites, our server hardware platform is single 1GHz P3 CPU, 512MB RAM, 20GB IDE disk. The software platform is a minimal SuSE Pro 9.3 (no GUI) with Apache 2.0.54 and Squint 0.3.16 for log monitoring. At each of the spoke sites, Squid was built from Squid-2.5.STABLE10-20050725.tar.gz using configure options: --enable-dmalloc --enable-storeio=aufs,ufs --enable-carp --enable-icmp --enable-delay-pools --enable-snmp --enable-err-languages=English --enable-default-err-language=English --disable-hostname-checks --enable-underscores --enable-stacktrace. The SQUID.CONF is basically the default settings with following exceptions: Cache-Dir is aufs size=1024 TopLevelDirs=16 and SecondLevelDirs=256 Cache_peer (Parent IP) parent 80 3130 round-robin Anyone have any good suggestions? Thanks in advance. Bob Morrison, CNE, MCSE Network Administrator Wallingford CT Public Schools
