Re: Soft raid 10

Dave Hall wrote:
> Just adding an annecdotal data point for the hardware option Steven mentioned
> he might consider if a software solution didn't perform good enough.
>
Ahhh, so you weren't specifically speaking about Steven's benchmarks, then?
> Computationally intensive but hardware RAID shouldn't be using any
> more CPU than a plain old disc controller.
No, but it could be slower, because it's dependant of the speed of the
processor on the RAID controller. I consistently see that on new
machines, Linux RAID5 far outperforms hardware RAID5 controllers, simply
because the processor doing the calculations is faster.

Therefor, it's not that surprising to see even a really fast single disk
outperform hardware RAID5 if the RAID card has a slow processor on it.

However, it's also still not valid to compare RAID5 (which is very slow
either on the controller or the CPU side because of hashing) with RAID10
(which is basically only limited by your disks and I/O busses).

Graeme
Received on Fri Jan 26 16:19:55 2007