Dear all we are using squid as Transparent proxy with following setup
1) Router Cisco 7200 series VXR IOS 12.3 with WCCP v1 enabled. 2) Red Hat Linux 9 with wccpv1 patch ( 4 boxes. 2 on each downlink ) Machine Hardware Intel Entry level Server motherboard Intel(R) Pentium(R) III CPU 1200MHz RAM 1 GB 3 SCSI disk on each system ( one for OS and 2 for cache) average 3000-3800 reqs/min at peak hours. HIT ratio is very low between 35-40 % squid cpu usage 70-85 % which is i think very low ? i have applied low lentency kernel patch on one of box and tested with Dbench ( got the following results before appling patch Throughput 87.6516 MB/sec 30 procs Throughput 101.827 MB/sec 30 procs Throughput 107.914 MB/sec 30 procs Throughput 119.217 MB/sec 30 procs Throughput 124.776 MB/sec 30 procs After appling patch Throughput 154.831 MB/sec 30 procs Throughput 151.445 MB/sec 30 procs Throughput 156.55 MB/sec 30 procs Throughput 149.864 MB/sec 30 procs Throughput 147.731 MB/sec 30 procs you can see the difference. and now running squid on same box i got around 50-52 % hit ratio i am thinking running squid on port 80 .so i can skip the iptables . any drawback running squid on port 80 ??? next i am thinking how to improve squid network I/O any improvment throughts and ideas are highly appreciated squid.conf is attachted below. Thanks and Regards UW my squid.conf is as follows; cache_mem 32 MB cache_swap_low 90 cache_swap_high 95 maximum_object_size 4096 KB cache_replacement_policy lru cache_dir aufs /cache 8192 16 256 cache_dir aufs /cache2 8192 16 256 cache_access_log /var/log/squid/access.log cache_log /var/log/squid/cache.log cache_store_log none __________________________________ Do you Yahoo!? New Yahoo! Photos - easier uploading and sharing. http://photos.yahoo.com/
