VMware made server virtualization all the rage recently, but what they do was done by storage folks a thousand years ago. Making one thing look like a bunch of other things is cool, but it's been done.
No matter what, it's a good idea to enable an underutilized asset to be chopped up into a bunch of virtual assets that can increase the overall utilization, decrease the management burden of dealing with multiple hunks of iron versus fewer, and centralize control functions wherever possible. So, why after 87 years of mastering virtualization technology at every layer of storage has it still not become even remotely mainstream – especially since the V word is no longer taboo thanks to VMware?
Standards. By the time VMware caught fire with the exact same value proposition storage virtualization has been touting for decades, servers were already commoditized. Intel won the processor war, Microsoft won the broad based OS war, and Linux became the de facto Unix. Dell, HP, IBM, etc. all have their variants of servers, but the difference is in packaging, not in the core technology. The market had already commoditized – so VMware didn't face any significant resistance from the server vendors. They didn't feel the need to defend their proprietary implementations nor fear that those implementations would be commoditized – it had already happened.
Storage isn't there yet. We barely have standard interfaces. If you can't even agree on how to talk to something, how will we agree on what it does?
It wasn't long ago when every server vendor also had their own chipset, O.S., and apps. How did that work out? Why do continue to permit storage vendors to dictate forced incompatibilities?
What it does tell me is that A: history repeats itself, B: history repeats itself again, C: history tells us that if this story is repeated again, that anyone who doesn't follow along with the inevitable standardization (and yes, commoditization) of storage systems will end up in the scrap heap of tech along with Wang, Prime, and DEC.
The answer will be forced, and will not appear willingly. It never does. It will end up being some de facto winner who brings the market all the scale-out, automated, self-everything next generation "infrastructure 2.0" stuff that virtualizes everything in its stack – without adding cost to the commodity underneath. Why? Because like Cisco, Adobe, or anyone else who has won by becoming a "free" standard bearer, the value they drag is monetized on top of the interface or base function. No one makes money on a protocol – they make money adding value on top of it.
IBM and HP were able to adapt, but most went to sleep with the switches. Storage guys can adapt, but will they? In 1981 my favorite Boston band was the Del Fuego's. They are old now. The singer, however, adapted – and damn if he isn't doing kids music and damn again if my kids don't love him (Dan Zanes). He makes way more dough now then he did then, I'm guessing.
Comments