init observations

[Stephen J. Smoogen]

On 11/16/05, Jeff Spaleta <jspaleta at gmail.com> wrote:
> On 11/16/05, Stephen J. Smoogen <smooge at gmail.com> wrote:
> > A lot of commodity disk-drives are sub-par on these areas. This means
> > that programs will slow down because you are spending a lot of time
> > waiting for the disk-drive to give you bits because it keeps plugging
> > up its cache or not able to get various items off disk as fast as
> > advertised.
>
> Is there some reasonably valid understanding of how big an impact this
> makes in the average and worst case? Sort of a measure of the spread
> of disk-drive performance out in the wild.

Not that I know of.. I am guessing the proper methodology would be to get
3 disks of the same model (to check for divergence inside a model)
1 set of 5000 RPM ???
1 set of 7200 RPM
1 set of 10000 RPM
1 set of 15000 RPM

and sets depending on disk cache sizes and such.

And then plug them into the same hardware system so that one should be
just testing the diskdrive. Once that divergence testing has been
done.. it would next to see what changing controllers do [that is the
second thing we subjectively saw as a bigger improvement than getting
someone a faster CPU.]




--
Stephen J Smoogen.
CSIRT/Linux System Administrator