[K12OSN] Are fast disks really that important and why?

[Calvin Dodge]

On 6/2/07, Sudev Barar <sbarar at gmail.com> wrote:
> On 03/06/07, Calvin Dodge <caldodge at gmail.com> wrote:
> [SNIP]
>
> Normally surveys mentioned would be looking at performance parameters
> for a single user desktops. Am I correct in this assumption? Since

No. The studies (there were two different studies, but Google was the
only one which stuck in my mind) were from server farms, not
single-user desktops. Both surveys were concerned with disk
reliability and lifespan, not performance. That was my point, when I
said:

> > reliable than others.  They _did_ say there was little practical
> > difference in reliability between SCSI and SATA drives.

> there is no link to study I can not give a informed opinion.

Sorry. Here they are.

Google's study results are at http://labs.google.com/papers/disk_failures.pdf.

My memory was a little off - the survey which showed little difference
in reliability between SCSI (and FC) drives and SATA drives was at
http://www.usenix.org/events/fast07/tech/schroeder/schroeder.pdf.  A
synopsis of this paper is at
http://www.usenix.org/events/fast07/tech/schroeder.html

Calvin