Matus UHLAR - fantomas wrote:
On 16.07.08 13:20, Rhino wrote:
Have Squid2.7STABLE1 running on Debian Linux box with 4 278 GB drives
allotted for caching.
System has 32 gig RAM, and serves approx. 10k users.
Seeking input as to the most effective config options on this system in
order to reduce latency
and maximize throughput.
In addition to what Hendrik said
- one cache_dir per disk
- you can skip logging if that's not needed (but I wouldn't do that)
- use one disk or maybe two mirrorred disks for system (logs) and the rest
for spool
- if you want to have more space, you can create smaller spool on the same
disk system it. If you mirror it, only mirror system partitions, don't put
spool on mirror (unless you don't want to loose any part of your spool
when a disk fails)
- using small(er) COSS spool on the system + logs disk could help
performance, if smaller files were put there and only bigger on other
spools) however COSS is not that stable yet (afaik).
COSS in 2.7 is pretty stable for files under 500KB. Its 3.x where the
stability fixes are still missing (small portage project for someone
interested ??).
Amos
--
Please use Squid 2.7.STABLE3 or 3.0.STABLE7
