This article is more than 1 year old
NetApp's all-flash FAS array is easily in the top, er, six SPC-1 machines
Better than 6th place, not quite good enough for 4th
NetApp has published an all-flash FAS8080 EX SPC-1 benchmark report [PDF]: 685,281 IOPS, $2.68/IOPS, and 1.23-millisecond response at 100 per cent load. This is the fifth-fastest SPC-1 result in our files in IOPS terms.
Top SPC-1 performance (IOPS) results with NetApp's FAS8080 EX in the fifth slot (click to enlarge)
It is also ranked sixteenth in terms of dollars per IOPS performance. However, this figure is always a pain to evaluate as vendors' list prices are often used, and those vary dramatically from the actual street prices.
The SPC-1 benchmark measures a storage array's ability to respond to IO requests. NetApp blogger Dimitris Krekoukias (aka Recovery Monkey) gives us his take on the result.
Read his blog to see his assertions about how NetApp used a realistic configuration, unlike other vendors, and see his ranking of systems based on their access latency.
Top SPC-1 $/IOPS results. NetApp's FAS8080 EX is ranked 16.
The comments on the blog post are interesting, with a ferocious Krekoukias firing off fierce replies about competing and higher-scoring SPC-1 benchmark results, for example:
The systems “above” it have only a tiny fraction of the FAS8080 functionality and two of them are either systems from a company that might just go away any time now or from a company that has no foothold in most of the world for various reasons.
He's referring to Kaminario (the biz that may just go away) and Huawei (the company with no foothold), by the way.
As for Hitachi Data Systems, the top-scoring vendor, Krekoukias says: "Only HDS has a decent offering and they barely passed the unused capacity limit. Probably to get the working set size small enough to fit in cache as much as possible."
SPC-1 system manufacturers tend to say the other vendors game the results. It's ironic that a benchmark intended to provide a standardized way to compare storage arrays should be the scene of disputatious arguments about how to interpret the results. ®
Biting the hand that feeds IT
