Should squid RAM management be left to the kernel or configured explicitly?
Howdie, fellow Debianites!
Looking for some input on the following dilemma:
I'm running a small home-LAN squid3 server with 1 GB disk cache and
trying to optimize its performance. The Lenny machine running it has
2.5 Gigs of RAM, of which more than 1.5 Gigs is being used as cache (as
shown by 'free -m'). Do you think there would be any point in
optimizing the squid.conf MEMORY CACHE OPTIONS (such as cache_mem,
maximum_object_size_in_memory and memory_replacement_policy), or would
it just "duplicate" the caching provided by the kernel? I mean, aren't
all those objects already being held *in RAM* anyway?
Disparate opinions welcome.
Regards, and have a nice day!
Certifiable Loonix User #481801