Re: [squid-users] Performance question

On 29.06 11:21, John Halfpenny wrote:
> what i +did+ say was that it beats a single ata in terms of general drive
> performance. you only have to perform a format to see (at a very basic
> level i grant you) that it is faster.

Well, format is something specific (for example because % of backup UX-like
FS' superblocks lowers with higher capacity, so one big filesystems has less
superblocks than two half-sized FS's), but we don't format that often to get
big benefit of speeding up this operation :)

> your comments about hw raid as not being particularly better are
> confusing, as hw raid on a scsi setup will almost certainly beat any ide
> configuration you care to mention. this isn't what i've done here, but if
> i had the money for a scsi raid card i wouldn't be digging out old kit to
> install squid onto. :)

You are comparing two uncomparable things. Of course, HW array with
faster (and SCSI) disks will work faster than slower (and IDE) disks.

What I want to say is: when we already do have some disks, it's more
effective to create filesystems on each of them and use them as separate
cache directories than playing with HW/SW stripping.

And we don't have to buy HW RAID card.

And if any of disks fails, we'll only loose part of our cache.

> the downside of a lost cache through raid0 is a risk that we have to face
> through lack of funds, and i'm prepared to mount another drive in place of
> the raid should it go down, which would only take a few minutes.

IT will take a few minutes even as a separate drive, and while you'll create
FS on it, SQUID can run without that particular cache_dir.

-- 
Matus UHLAR - fantomas, uhlar@fantomas.sk ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
"Two words: Windows survives." - Craig Mundie, Microsoft senior strategist
"So does syphillis. Good thing we have penicillin." - Matthew Alton