In a recent interview with CIO Insight, Kenneth Brill of the Uptime Institute posits that the growth of demand for data processing is likely to cause enterprises to repeat data center construction projects every four to eight years. What?! Do it again?
In this discussion, Mr. Brill is reflecting on the way that our decisions about scaling infrastructure (and facilities) have followed an economic model that isn’t well aligned with the pragmatic realities of the way costs associated with operating the facilities and the IT footprint impact the profit-and-loss statements. That is to say, the focus often tends to be on horsepower per square foot, and the impact from utilities costs are often not properly represented in the decision making process.
Without going into elaboration of that dialogue, I’ll second the argument and agree that this certainly has been the case. I would think that the four-to-eight year rebuild cycle (or whatever the proper rebuild cycle is) is dependent upon the vertical in which the Business operates, but the point is well taken. Mr. Brill suggests that this is an economic crisis that is manifesting in a trend to shift costs of the data center to the IT organization. I’d like to add some additional perspective to this discussion, focusing on the dilemma I observe with my Clients trying to win funding approval for data center projects.
Data centers are not only very expensive to operate, but they’re expensive to build as well (excuse both understatements). In the enterprise space much of the current building boom is due to either a desire for facility-driven availability improvements, data processing footprint that demands more MEP capacity than current facilities deliver, Business growth, or a combination of the above. In most cases, this is not a project that was planned very far in advance. It’s often a surprise (“Hey, we’re running out of cooling capacity”). The costs of these projects can run from $20M on the low end and on up into eight figures (‚Äòlet alone those monster projects over the past year that are investing $billions in data center construction).
Let’s take a run-of-the-mill example. Let’s say our company is planning to build a 20,000 square foot, 3 Megawatt, Tier-3 data center. With those vague requirements, a reasonable shoot-from-the-hip cost bogie is $50M to build this facility. Do you want to justify a $50M capital project to your board? There are more enjoyable things to do for sure. Do you want to try it twice?
Going back then, to the comment about needing to build a new data center every four to eight years, even if we manage to afford it once in this career it’s just not reasonable to think that we can afford such a capital expenditure over and over unless our Business is blessed with a long term explosion of margin. Even if we could, we surely don’t want to. In a somewhat circular way, this is confirming Kenneth Brill’s suggesting that what we have here is an economic crisis for the Business.
So looking ahead (as we always do) one has to ask, “How can I give this problem away?” This is in fact, one of the opportunities in the Cloud Computing trend. If we can leverage the cost of data center facilities across a number of enterprises maybe we can find a more favorable financial model to support the scaling of our business.
Some firms will find this more palatable than others. One degree of variable in this notion is the maturity of services available from a cloud platform to begin with. If our application is sufficiently agile that any color of MIPs and Megabytes will do, or if we use a service that’s sufficiently commoditized and available by now in the cloud then we’re in good shape. If we have a tangled web of legacy applications, we’re probably out of luck until our application infrastructure evolves to something more portable (even outsourcing is difficult to pull off successfully in this case). And then there’s the fact that even if you find cloud-based services appropriate for your enterprise, there is a cost of changing from what you have today to what you’ll use tomorrow in the cloud. In this last case, new businesses potentially have a distinct advantage as they can craft their assortment of services from those available in the cloud.
Now it’s one thing to ponder these problems in isolation, but the fact is that in almost all cases, there are multiple concurrent strategic trajectories in play in the enterprise. At any moment, we may have the operating model of the company shifting from one profile to another, we may have development of a new enterprise architecture, and we may have an effort to mature the IT organization in general. The success of one trajectory is often mutually dependent upon another. For example, the benefits of data center consolidation may not be fully realized until I can standardize platforms in the consolidated facility. I may not be able to standardize platforms until business processes or customer data converge.
Because the data center is arguably the foundation of the data processing footprint for the enterprise, the role of the data center in the financial and IT plans is best supported if linked to the holistic view of the strategic intent of the Business, and the concurrent trajectories defining the roadmap to those goals. This is more easily done in some organizations than others.