Bunging servers in disk arrays achieves nothing. There, I said it
If you disagree with the Reg storage desk, explain yourself
Blocks and Files A while ago in-array compute was going to be a big thing, with apps running inside VMAX and VNX arrays using spare controller engines and getting rid of network-lagged data access latency.
DataDirect Networks went down the same development avenue, and Lustre and the GPFS stack certified to run in its storage arrays.
Bringing storage and compute closer together is being implemented, but in the reverse direction: storage is being added to compute.
The first way of doing this was to have the networked storage attached to an extended external PCIe bus, or to an InfiniBand link; thus becoming a local SAN with no Fibre Channel or Ethernet/iSCSI access lag.
It was not popular.
The second way, an indirect one, is to have flash caches in servers with the contents managed in some way by a connected storage array. It's a halfway house, as a third way is to have a local flash array connected to servers by InfiniBand or fast Ethernet. This is coming, and EMC's XtremIO technology should help usher it in.
The fourth way is to have flash memory in servers treated by the host OS as an adjunct of main memory, called storage memory. Effectively, this would get in-memory processing of an application and its data set. This software-led approach is being promoted by Fusion-io and Virident.
Server vendors like this server-centric approach of bringing application data and compute closer together. So do customers. They buy servers to run apps and want those apps to run faster. If networked storage takes too long to access, then bring it closer to the servers.
Don't start putting servers in storage arrays; that's like putting a car's engine inside its petrol tank. If fuel takes too long to get to the engine, you bring the petrol tank closer to the engine. Simple.
In-array compute is a dead duck that isn't going to fly, except in small market niches where buyers are array-centric in their approach rather than server-centric. To mix the metaphors, in-storage processing is for the birds.
If you agree or disagree stick a comment in the forum topic on this subject. ®
Sponsored: Hyper-scale data management