On Tue, 18 Nov 2003, Philip Choy wrote:
> Expected # of users of these cache: 1000, and currently 2 pilot users.
>
> The specification that I have :-
> 1xE1 2mbps connection
> 2GB Physical RAM
> 2GHz Xeon
> RHL 9.0
> squid 2.5.1
Ok, but if you are installing a new Squid why are you not using the
current STABLE version?
> /var/spool/squid mounted as default ext3 on 1 logical vol of phy vol group
> of 4 RAID-0 SCSI disks of 20GB.
Ok, but it is easier to adminstrate if you mount each drive spindle
separately, and refer to them by separate cache_dir lines.
> cache_mem 768 MB
Why this very large amount of cache_mem?
> cache_dir ufs /var/spool/squid 32000 64 256
Your really want to use one of the async cache_dir types if you are
looking for any kind of performance (aufs or diskd, aufs generally
preferred on Linux)
Regards
Henrik
Received on Tue Nov 18 2003 - 08:55:19 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:21:21 MST