December 9, 2014
I love the movie Shane. This George Stevens’ classic stars Alan Ladd as Shane, an aging gunfighter seeking to live his life within society rather than on its periphery. Unfortunately, despite his efforts, he soon finds that a man can’t change who he is, and, after saving the homesteaders who have accepted him as their own from a violent rancher and his band of hired guns, the movie ends with Shane slowly riding away as the plaintiff cries from little Joe to “come back” fade in the distance. If you haven’t seen it, let me just say this is some pretty moving stuff, and I’m getting a little misty just writing this. Shane’s inability to adapt to a changing world reminds me a little of the predicted fate of data center expressed recently by some industry followers. Both IDC and Cisco have released predictions regarding the future of the data center in the coming years, and to listen to them, someone from central casting better start looking for the Alan Ladd of the digital age.
Although they differ somewhat as to the driving forces and timeframes, both organizations project the data center going the way of the gunfighter in the coming years. Cisco posits that the cloud will limit future data center growth and predicts that by 2018 more than 75% of data center workloads will be running in cloud data centers. Not to be outdone, IDC has decided to “see” Cisco’s prediction and “raise them” by pegging 2017 as the zenith of data center growth with a slow decline thereafter. Placing the blame squarely on the changing role of the corporate data center, the prognosticators at IDC see future corporate computing needs as becoming too complex, manpower intensive and expensive for most company’s to continue to operate their own facilities.
Interestingly, both organizations feel that this tectonic shift in the industry will give rise to fewer, but bigger data centers. While it’s easy to become enamored with the gigantic facilities built by the Googles, Facebooks and Apples of the world, I’m not so sure that in the future data centers will all be the size of small planets, and IDC admits that that all types of data centers except for on-premise server rooms and closets—which weren’t exactly two high growth areas top begin with– will continue to grow. While I believe that consolidation and the elimination of older facilities (in many instance, “old” will probably include sites that have been around for as few as five years) will reduce the number of data centers in absolute numbers, the need for data floor space will continue well past the 2017, 2018 milestone estimates that IDC and Cisco have offered.
Peering into the future is always somewhat of a murky business. Some predicted events never come to fruition, some do, while others do but not quite in the way that they are envisioned. I think the latter category is the most likely for the industry. Like most things, the architecture of data center networks is not a “one size fits all” proposition. I think that the high volumes of data and the increasing speed in which people want to access it, will result in more distributed structures with large centralized data centers working in conjunction with multiple edge locations, and, in some instances even micro facilities. Thus, while Cisco and IDC may be technically correct, folks are still going to be building a lot of data centers for quite a while.
This latest evolution in data centers can also be expected to lead to a more stratified industry structure that will provide clarity for industry followers and end customers alike. If the history of other industries, networking hardware for example, is any indicator we should expect to see consolidation amongst the providers of “massive” solutions. By definition the shear size of these sites will have an inverse relationship to the physical number of data centers built. Thus, a number of existing players will find their most profitable strategy to sell their facilities to a larger competitor and exit the business. Edge data centers will become the domain of a few smaller, nimbler firms that can quickly deliver a reliable product with all the features that a firm would include in a site if they built it themselves, with the ability to quickly scale up to meet demand in line with corporate budget and capital requirements. At the present time, the need and requirements for “micro-sites” of 250kW or less requiring little, if any, physical support remain a work in progress. The continued evolution of software defined storage and server functionality can be expected to be the primary drivers of innovation within this space.
For some, the recent predictions from IDC and Cisco may be cause for concern, but isn’t this really more of an alarmist reaction? Every industry changes of over time, and in businesses like ours where the “old days” means two years ago, these projected shifts in the nature of data centers should be viewed as the catalyst for new innovation and product direction that they are. In other words, were may be nearing the end of the first act, but were a long way from watching Alan Ladd ride off into the sunset.