Re: [squid-users] Number of Spindles

Nyamul Hassan wrote:
> Hi,
>
> I was reading through Duane Wessel's book, and I must say, it is a GREAT
> book. If only I had read it earlier, my life would've been so much
> better! :)
>
> I was reading through Appendix D: Filesystem Performance benchmarks,
> where Duane says the following:
>
> "The primary purpose of these tests is to show that Squid's performance
> doesn't increase in proportion to the number of disk drives. Excluding
> other factors, you may be able to get better performance from three
> systems with one disk drive each, rather than a single system with three
> drives."
>
> Isn't this contrary to what we've been seeing over and over again in the
> forum, that increasing the number of spindles distributes disk IO, and
> increases cache efficiency by avoiding io_wait for the CPU? Or is it
> that, when you use SCSI U320 HDDs, your io_wait becomes a non issue, and
> that is what Duane was referring to when he started the concluding line
> with "Excluding other factors"?
>
> Could the gurus in this forum please shed some light on this?

The books is rather old. From the days of Squid-2.5.

I think diskd was developed in Squid as a solution to that problem, then
AUFS came along to improve things even further as a native OS layer.

Plain old single-threaded UFS still displays many of the performance
issues mentioned.

Amos

-- 
Please be using
   Current Stable Squid 2.7.STABLE5 or 3.0.STABLE10
   Current Beta Squid 3.1.0.2