Hi,
I remember now where I got the idea that under one-disk-down raid-5 was actually
faster than with all disks available. It was at a talk by the guy from Data
General who has a *lot* to do with their Clariion disk arrays. Here's what I
remember:
1) overall read performance is (as rightly pointed out) *slower* - 1 in "n"
reads require the data (from the block on the missing disk) to be reconstructed.
2) overall write performance is *faster*. Why? Because 1 in "n" writes no longer
requires the parity calculation because there's nowhere to store it - the parity
block is on the missing disk.
At least I think that's right :-) If it isn't, I'm going to hunt down the paper
from that talk and read it again.
Colin
-- Colin Campbell Unix Support/Postmaster/Hostmaster CITEC +61 7 3227 6334Received on Thu May 01 2003 - 19:37:27 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:16:09 MST