Optimal for single session? Maybe. Isn't this roughly what RAID-2 was supposed to do? Of course, even then, it's really only optimal if the platters all spin and the heads all move synchronously, isn't it?
I was under the impression that this is why RAID-2 is rarely implemented. (Normally, I'd say "never implemented", but on this list, I just *know* somebody would correct me...)
I don't know, really. Probably for single-session applications, especially those doing huge sequential I/O, you'd do quite nicely with just two or three of those "Stinking Huge (TM)" ATA disks that make us shudder whenever our PHBs mention them.
;-)
Sadly, I've never bothered to benchmark this configuration, but I *am* pretty confident that it will be far less-than-optimal for most concurrency levels much greater than 1...
Comments inline.
On 5/25/06, Mark Brinsmead <pythianbrinsmead@xxxxxxxxx> wrote:
> > (*) Make your RAID stripe width really narrow (like maybe 128 or 512 > bytes) and then > make your database blocks large (like 8KB or 16KB) and do lots of > multi-block I/O, > so that every I/O is assured to engage every physical drive. (Kiss > concurrency goodbye > and watch your RIOPs numbers plummet) >
Seems like somthing like this might be optimal for single session. Benchmarks? ;)
-- Cheers, -- Mark Brinsmead Staff DBA, The Pythian Group http://www.pythian.com/blogs