[SQU] Performance optimization with diskd

Hello.
My main cache is getting pretty busy, with .5M hits/hour of peak traffic.
I am thinking of ways to improve (even if marginally) performance.

This is my latest thinko:
I'm using diskd, JBOD, one cache_dir per physical disk.
If I were to put on each physical disk two (or more) cache_dirs
half the size of the current one, I'd potentially have
multiple pending IO requests per disk and thus be able to
exploit OS-level parallelism, elevator algorithms and whatever
else.

Is this brainless blabber or does it actually make some sense?

-- 
	/kinkie
--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html