[Gllug] Poor disk performance on IDE raid

[Simon Wilcox]

On Fri, 3 Jun 2005, Martin A. Brooks wrote:

> Not wishing to be harsh, but if your system relies on low latency I/O
> then you have barely picked a worse setup.

I agree but these are nodes in a web cluster so it's not critical, I was
just a little surprised to see it being quite so slow.

> If you want low latency with RAID then you need more spindles.

Indeed, I often have trouble explaining to the $client why it is better to
use more smaller disks than fewer big ones.

Simon.

-- 
"I'm not happy with my wash!"

-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug