Hey, Depends on what is purpose of the proxy. if it's only filtering then just use SMP without HDD case. I assume this is not the case. In the old days you would just run couple instances of squid and intercept by a RR like into different squid instances. In squid SMP you can try to use ROCK store that stores small objects in order to use SMP and cache_dir together. the first thing is to just run SMP without any cache just to start it up and make sure it runs smoothly.. In the next step you can try to add the cache_dir. If you want to measure the RAID1 efficiency feel free. Dont rush into SSD since it's nice but you now have the specific hardware that can work fine just like that. What Raid are you using?? HW or SW or LVM? I assume that you are not only trying to save couple small pieces of bandwidth?? Eliezer On 07/17/2013 11:44 AM, Travel Factory S.r.l. wrote: > > I already asked a similar question about 1 year ago but at that time we > had to use the servers for other jobs. > > Now I've got brand new servers. They are very powerfull, 16 2.90ghz > cores, 32 GB ram, 3 couples of raid1 146GB 15k disks and SSD may arrive > in the near future. > > I installed 3.3.5 and yesterday upgraded to 3.3.8. > > There is a single squid instance on each server, serving about 1500 > users with a proxy.pac splitting requests. > and as you may imagine load is very low and sometimes "top" has more cpu > % than "squid". > > I'd like to know which kind of configuration setup is possible to use > all this work power and ram. I read about SMP or about multiple squid > instances running on different ports... > > Which settings to put ram usage at maximum ? > > Thanks > >
