Re: Squid cache size? (was Re: Linux firewall vs Windows and Hardware based firewalls)
On Mon, 2003-08-04 at 06:32, Paul Johnson wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Mon, Aug 04, 2003 at 10:49:47AM +0100, Karsten M. Self wrote:
> > Incidentally, how much space are you dedicating to your Squid cache? I
> > know that this can't be assigned directly, so either the segment size,
> > or net use (du -s on cache) would be useful.
>
> A gigabyte does pretty nicely for my users.
Being a cachemaster (well formerly at least). I had 30GB of disk space
on U160 SCSI 5 Channel Raid controller using RAID 5 and LVM.
I used ext3 with no ATIME being updated.
@30GB of cache, it needs ~ 1GB for the hot-indexing of the cache.
I also devoted 1GB of Memory for Hot-Serving of cache items. My maximum
object size was 120MB or just about the size of most Large patch sets
from Microsoft, Novell, IBM, etc...
This particular cache served ~2200 users concurrently... had an average
cache hit rate of 40%... which meant ~ 40% saving in bandwidth
requirements. So at the 10Mbit/sec pipe we had... it "felt like" ~
14Mbit/sec.
This cached HTTP and FTP. It helped out tremendously.
--
greg, greg@gregfolkert.net
REMEMBER ED CURRY! http://www.iwethey.org/ed_curry
Flies dance operas to your wisdom.
Reply to: