> Can anyone explain why a software raid5 array of N disks has
> significantly lower read performance than a raid0 array of N-1 disks?
> I'm only considering the case where there are no drive failures.
My guess would be that since the parity is striped (unlike raid4),
that you'd actually have to figure up which drives have the data and
which one to skip, instead of being able to blindly loop around all
drives in the array like raid0. This is a WAG at best, though :)
Pure read performance shouldn't suffer with parity calcs unless
the array is rebuilding... or is there some case where it'd recalc
the parity and rewrite it back out regardless? Hmmm
James Manning
--
Miscellaneous Engineer --- IBM Netfinity Performance Development
