[Beowulf] GPFS on Linux (x86)

[Mark Hahn]

> If someone would be so kind as to help me find *real* data that
> demonstrates higher SATA/IDE failure rates as compared to SCSI, I would
> most appreciate it.

I have only two very wimpy factoids to offer: my 70TB HP SFS
disk array (36 SFS20 shelves with 11x 250G SATA each) has 
had just one bad disk since it was installed (say, March).
so that's one disk in 1.7Mhours, aggregated, actually a lower rate 
than I would have expected...

that storage is attached to 768 compute nodes, each of which has 
2x80G SATA, which I believe have had no failures (6.6Mhours!).

the hardware was all assembled and burned in quite a while
before being delivered, so this should be the bottom of the bathtub.

during some periods, machineroom temperatures have not been 
exceptionally well-regulated :(