Don't let me down here people...
1) A RAID is NOT just a disk subsystem.
2) It's not "one big mutha of a hard drive"
It's not like sticking a big bunch of disks on the end of your Firewire bus, my friend. If it was, it would be a LOT cheaper and you'd be thinking of suicide when a disk blew up and you lost all of your data.
RAID is chosen for:
-- Speed (you can get 400MB per
second transfer speed off some RAIDs!!)
-- Storage capacity (obviously)
-- Less nearline or offline data requirements (tape drives, jukeboxes, DVDs, etc.)
-- Probably a load of other reasons that I don't know about
A RAID isn't just a dumb load of disks. It has one or more controllers that manage the RAID subsystem independently from the server, presenting the disks (in whatever config) as a single "disk" (or more rightly, "volume") per RAID array.
OS X has software RAID built-in to the OS, but no-one with half a brain would EVER touch a software RAID solution - it's just not reliable enough. They also take up system memory, and eat clock cycles. Being in competition with whatever apps are on the box in the first place, software-based RAIDs run dog-slow compared to hardware-based ones.
Hardware-based RAIDs are true multi-tasking machines that do not occupy any host system memory and are not OS dependent. But the main reason that they're better is that hardware RAIDs are highly fault-tolerant. If you're putting all your data in one place, it better NEVER fall over....
The question is: what sort of controllers are they using? Clock speed is almost irrelevant, as the Controller is built as a "hardware array" that doesn't HAVE any software to run on it.
XRAID attaches to a SERVER (XServe, of course). All the other machines access the data off this main server via networks (Ethernet, Gigabit, SAN, whatever).
That's how it works. Hope I've helped.
BTW, just to confuse everyone a bit more, a disk storage repository without a dedicated RAID controller is called a JBOD ("Just A Bunch Of Disks")
