The results are what I would expect from the Gen 3 system, and SPC-2 benchmarks do play up on the strengths of the XIV architecture. SPC-1 results would be desirable to see, especially once the new 7.5TB read cache is incorporated into the system.
In my own testing I’ve managed to garner 1.3GBPS throughput utilizing IOmeter scripts and virtual machines. I belive direct connected hosts without the additional overhead would perform better. At some point in the future I will be workign to post some swingbench testing results.
The Gen3 model is a vast improvement over the Gen2 in terms of design. Gone are the 1Gb interconnects in favor of the superior infiniband. Near-line SAS over SATA in a 2TB size, 360GB of cache, 24 8GB FC versus the 4GB of Gen2, and 22 1Gb iSCSI ports. Upcoming changes this year will see the addition of 7.5TB of read only cache! Total useable capacity for a full frame is 161TB versus the Gen2 79TB. I’m guessing that the IO/Data nodes are X-Series 3690’s with dual nehalem 8 cores and 24GB of RAM per node.
The unit is deeply discounted at 70%, if I had the need for the horsepower, I would definatley jump on a trade-in. I’m sure IBM leasing would dangle some attractive packages to get these units moving. It’s nice to see list prices avaialbel to the public, this is something that bothers me about storage systems because getting list pricing can be a daunting task.
I’d recommend perusing through Anthony Vanderwert’s site for a plethora of XIV related info and posts. I’ll be putting up more later.