We have two caches both getting approx 750000 tcp requests/day.
Sparc 20, 448Mb ram.
Squid 1.1.16
Cache_swap 11000.
Cache_mem 32
Stable squid size approx 310Mb. [uptime of 16 days now]
Ultra 140, 256Mb ram.
Squid 1.1.17
Cache_swap 7500.
Cache_mem 24
Stable squid size > 230Mb [I have to restart it when it starts swapping -
about once every three days at present :(]
Performance is _markedly_ degraded when the cache starts swapping.
Average request service time goes from approx 3 seconds to at times over
10.
Annoyingly when the object database has been reloaded after a restart the
size of squid in memory is approx 50% of the stable size, but tends
towards the stable size quite quickly.
John Sloan
Systems Administrator
UUNET UK
Received on Thu Oct 30 1997 - 09:49:42 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:22 MST