I manage a cache for about 250 users. Pretty much all the settings are the defaults. We're running it on a dual-Xeon machine with 8 GB of RAM, so performance is never an issue. We even use SquidClamAV to scan every downloaded object for viruses with no obvious slow downs. We also use a lot of ACLs, including the entire "malware block list." All told there are over 20,000 separate ACLs. Squid still never misses a beat. The machine also performs a number of other tasks like scanning inbound mail with MailScanner and hosting my client's DNS and web sites.
The only things I changed from the default (CentOS) configuration were expanding the cache size and permitting more file descriptors:
Code:
cache_dir ufs /var/spool/squid 10000 16 256
max_filedesc 16384
A search for "optimizing squid" brings up a lot of articles.
Bookmarks