There have been suggestions recently that while enterprise flash storage will undoubtedly have its place, it will not become dominant, as others have predicted. However, past and future lessons alike suggest that we shouldn’t be too skeptical about the rise of all-flash Solid State Disks (SSDs) in the enterprise.
It is natural for most of us to resist change — or what we see as excessive change. We worked hard to get where we are today, and “if it ain’t broken, don’t fix it” remains a useful motto in enterprise IT. The problem is that we cannot stop or control change, no matter how scary it may be.
That is part of what makes Doctor Who’s silver foes the Cybermen — and their Star Trek cousins the Borg — so frightening: with their hyper-evolutionary ability to adapt to defeat weapons used against them, they are about the fearsome speed of technological change, as well as the more obvious fears of losing your identity and emotions, and being forcibly subsumed into “the machine”.
It is the same fears and lack of understanding that leads old industries to discount the new, assuming that although they too were once hot young innovators, now that they have achieved greatness they are safe from further challenges. They mock new technologies for their failings, even though those failings are exactly what’s impelling the start-ups and others to redouble their efforts.
So Kodak and the other camera companies thought film’s higher quality would survive the arrival of the digital camera. (The ironies were that the first known digital camera was built in Kodak’s lab, and that George Eastman’s film had also been a hugely disruptive innovation a century earlier, when photography had its equivalent of the silicon revolution. ) Similarly, traditional metal-bashers pooh-poohed 3D printing and other additive manufacturing technologies, just as skeptics have more recently highlighted the likely ineffectiveness of printed guns as anything more than a frightener.
And now elements of the storage industry committed to hard disk and hybrid hard disk/SSD arrays are reporting customer concerns over consumer-grade storage and cost effectiveness. Indeed, in a recent survey for disk array builder X-IO, the majority of the UK IT decision makers questioned said they simply did not need and could not make use of the massive performance increases offered by SSD.
The thing is, all those criticisms were and are right: yes, early digital cameras were pretty poor; yes, you probably can build a better zip-gun or “Saturday night special” in any car repair shop; and yes, hard disk storage is cheaper than SSD and is still a better fit for most of today’s enterprise applications.
But they’re only right until the newcomers perform a Borg-style upgrade, which they undoubtedly will — and fast. Not only are digital cameras now the norm, but 3D printing has already become the standard for prototyping and short-run manufacture.
Similarly, although consumer-grade flash does indeed achieve its low price by making compromises elsewhere, the manufacturers have already responded to concerns by developing the likes of EMLC — Enterprise-grade flash which saves space and cost by storing Multiple Levels (bits) per Cell, as consumer-grade flash does, but is designed for lower error rates. And organizations such as SNIA, with its Solid State Storage Initiative (SSSI) Workload I/O Capture Program are working to make flash more relevant to contemporary enterprise workloads.
For IT, it will be a matter of timing: staying with cost-effective older technology just as long as it best suits our specific needs, but being ready to jump when the opportunity arises. So while flash’s disruptive and fast-learning newcomers will be almost as adept as the Cybermen and Borg at destroying empires, the rest of us are unlikely to end up being assimilated into a brainwashed Cyberian future. Well, not forcibly, anyhow.