As previously noted, I have an interest in running large-scale services on equipment which would be called underpowered by many, so I find that this 256-CPU-cluster-in-a-rack is in many ways a dream machine. Of course, I don't think of them as underpowered but as efficient for the task, scalable and finally, green.
I would go as far as to say that if there need be clouds, these are the type of infrastructure they should run on (of course, highly geographically distributed - maybe 1 rack per 10,000 people). These machines have the following characteristics:
- Up to 256 dual-core Atom CPUs with Hyperthreading in a 10 U rack enclosure
- Up to 64 blade-like "compute cards"
- Up to 64 disk drives
- The "compute cards" share some of the infrastructure in a semi-virtualized way, like the NICs and the drives
With advanced server-on-demand (i.e. automatic turn-on and turn-off) capabilities (which I don't know if this product has), this could provide very elastic computing power for distributed applications.