For years in mid-range primary block storage, architectures were much the same. Two controllers, active/passive or active/active, loops of backend connectivity connected to boxes of disk. And performance, so the vendors would tell you, was much the same.
After all, disk was disk, front-end and back-end connectivity was the same across all vendors, no one cared about the storage processors, so performance was an untouted metric. Yes, you’d see occasional benchmark submissions, but mainly to fill in a box for a RFP.
For the most part vendors sold on capacity or other features, not performance. They’d ensure performance was adequate, but that was about it. Products were sold largely like cargo vans.
The fact of the matter was that storage performance wasn’t increasing.
But over the past few years we’ve seen a change, enabled by various memory technologies and architectures. For example, server-side flash + SSD + tiering + scale-out = explosive performance. And now a few vendors see performance as a disruptive ground to drive market share.
The newcomers in the market are now leading with performance, not mentioning it in the footnotes. Let me offer some quotes from different vendors:
High performance storage systems with guaranteed quality of service…
You shouldn’t have to compromise on performance…
<Blank> is the most intelligent and efficient solid state storage platform to enable enterprises and datacenters to solve storage performance…
It will be interesting to see if the new vendors can make that message stick. Certainly some of the established players have been layering performance features on their storage for some time (FAST cache comes to mind), but will customers buy into that approach, or will their SLAs require a architectural sea-change to enable disruptive performance? Time will tell…