Dog walkers, the San Andreas fault ... and the storage industry

Hot lava welling up in the San Andreas fault between servers and storage

The mainstream array giants are toast

What are the effects of all this on the mainstream drive array vendors in the general enterprise market?

There is no primary data-storing future for networked disk drive arrays, none at all. That role is toast. And unless their suppliers can beat the hybrid/flash array upstarts, the mainstream array suppliers are toast too; buy them, beat them or lose to them.

Old-style networked arrays are being reinvented as hybrid flash/disk and all-flash invocations of their original designs. But they are not yet denting the fortunes of lower-priced invaders like hybrid and flash arrays from Nimble, Tegile, and Tintri and new-design flash arrays from EMC, IBM, Kaminario, Pure, and SolidFire.

Yes, they have hold-the-fort tech, good stuff too, but how long can they hold the fort with it?

Another threat is represented by SW-only/SW-defined storage, enterprise-class SW storage, used for bulk secondary data stores. It means the big storage battalions should start trembling, because once-global 5000-class CIOs start to think that Nexenta/Nutanix-class SW with SuperMicro-class hardware, high availability, great support, and broad-spectrum functionality is enterprise class, then they ask simple questions.

If all that effectively separates Nutanix/Nexenta/SuperMicro-type storage from EMC/HP/NetApp/whomever is a bezel design, then why should I pay the bezel tax?

Why should I, they ask, put my non-primary, non-mission-critical data on arrays priced for holding primary, mission-critical data, when that data is moving to separate flash stores, leaving secondary data behind that only needs cheap and deep storage?

Once CIOs start thinking this way, then when leases, depreciation periods, and maintenance contracts come to an end, they could migrate away from mainstream suppliers' arrays. There could be a sudden and brutal switch with revenues leaking away from the big six mainstream array suppliers, like a dam break with pent up business flooding out to smaller suppliers.

There are literally billions of dollars of installed base value at risk here.

Blurred lines

Overall, new storage lava is welling up into the server-storage data latency gap, filling it so that where servers stop and storage starts is no longer clear; the boundaries are blurring and overlapping.

This wholesale assault on the server-networked drive array latency gap, and mainstream drive array cost, complexity, data lake inadequacy and lock-in, makes this one of the best times ever to be around in the history of IT storage.

IBM 350 RAMAC disk drive

It’s as potentially transformative as the invention of the Winchester disk by IBM all those years ago, and Toshiba’s invention of flash.

Hindsight will tell us if this is true. But for now we can bask in the incredibly inventive world of new storage technologies that are revolutionizing how data is accessed, and giving multi-core, multi-socket, virtualized, containerized servers the data they need fast enough to get them running more and more application, and apps running faster too. The lagging storage IO-rate is catching up with compute’s progress.

It is a great time to be alive in storage. Never better. ®

Disclosure: This article is based on a set of notes for my talk given at Nexenta's openSDx meeting in San Francisco, US, last week. Nexenta covered my travel costs between the UK.