RSSDeath throes of the corporate data center
Well, they're not quite here yet, and I am not so sure they ever will be, regardless of what Dion Hinchcliffe thinks.
It may come as some surprise to anyone who has noted my longstanding and occasionally excessive cheerleading of cloud-based utility computing as the best alternative to expensive, complex, and inefficient corporate data centers, but although Dion's well-written and reasoned article embodies many beliefs I hold dear, I think it's overly optimistic. There will be corporate data centers as long as there are corporations; and it won't simply be sheer bloody-mindedness on the part of CIOs that keeps them that way. There are good reasons for some businesses to run their own data centers, and it will continue to be in their best interest to do so in some cases.
The economies of scale introduced by utility computing services are inarguable, as they are simply upsized from the same logical economics that make any data center an efficient proposition. Economics, of course, are not the only factor in selecting a data processing infrastructure, but utility computing combines them with flexibility, nullifying a traditionally adverse relationship between the two; while it used to be that the larger you got, the more systematic your management had to be, the abstraction layer of cloud computing largely removes that restriction from the corporate subscriber. It is this factor, if you had to pick just one, that makes the concept a feasible large-scale, broadly applicable way to centralize processing resources where previous efforts in that direction have been unsatisfactory in some respect.
While I buy into all this, I also know that there are unknowns. While I believe that everything that Hinchcliffe says is generally correct (and he is not advocating a mindless leap into these services by any means, but suggesting a methodical, practical approach which is entirely in keeping with the idea that this is an inevitable, logical progression), I am also wary of such generalizations. I think it's a no-brainer that some major businesses will retain their internal data centers for reasons of security, regardless of how secure cloud services become (and I am still waiting for the first significant exploits to surface in some major cloud provider; no one is talking about it, but start your stopwatch, it's coming) and Hinchcliffe might concede that, regardless of how logical it may or may not be. But I think there are other reasons that the traditional data center will survive, as well, reasons which are not so easily articulated. They have to do with the cyclical nature of technology and the diminishing, but real, fact of technical competitive advantage.
Hinchcliffe accurately predicts, I believe, the slow adoption of utility computing based on traditional (and prudent) hesitancy in some corporate circles, as well as from the inevitable delays of porting applications over and from the fact that a datacenter in the hand may, for some time, continue to be worth more than two cloud computing providers in a bush: if you already built it, you may find that maintenance and operating costs will for some years continue to be cheaper than your transition costs.
But I believe that other firms will retain their data centers for reasons of competitive advantage. I can't yet see what exactly those reasons might be, and indeed it's hard to argue that a costly, inflexible internal resource is going to benefit a company more than a scalable, flexible, cheaper external resource. Nonetheless, it's possible that in some business models, in fact it will. We have seen the industry move inexorably back and forth between distributed and centralized processing in its various eras, and I think it's short-sighted to suggest that this is the last of those transitions. There are advantages accruing to some early-adopters of cloud services now; is it so unreasonable to assume that there may in the future be certain advantages to those foreseeing and moving early to the opposite swing of the pendulum? Utility computing is often seen as evidence of the commoditization of technology on a grand scale, and a sign of the disappearance of competitive advantages from technology. And for the most part I agree with this vision. However, there are always exceptions, and I believe that these exceptions, however rare, mean that the corporate data center is far from dead.
