Arriving with a satisfying thud, Atrato’s Velocity1000 leaves a lot of detailed questions unanswered but also poses a re-evaluation suggestion to supplier chief technology officers. Is spindle multiplication the way to go to get performance increases and reduced RAID rebuild times?
Seagate has been putting the spindle increase view about but drive array manufacturers haven’t yet responded. Perhaps Atrato is the first sign of a changeover wave to 2.5-inch technology, a drive format downsizing move.
Another question prompted by Attrato is I/O itself. Is the way to get more I/O out of a drive array enclosure to increase the number of ports or to move to a single large pipe, an InfiniBand one say? Atrato doesn’t say which it uses, or indeed whether it uses Fibre Channel or Ethernet. (It has been asked but hasn’t responded yet.)
The idea of ‘fail in place’ has been seen before; IBM’s IceCube research project at Almaden comes to mind. It would be interesting to know if any of the Atrato peple had been involved with that.
It’s possible that 2.5-inch drives could be used to increase the capacity possible per rack enclosure. Also, if Atrato does use 2.5-inch drives, then why wouldn’t a further format downsizing increase I/O even further as even more spindles could contribute?
Lots of questions with few answers. If Atrato’s products do prove popular and if other companies begin to follow suit then the 3.5-inch drive form factor’s universal use in disk arrays could be about to change.