August 21, 2013
“Is this a dagger I hold before me? The handle towards my hand?”
In the works of the Bard, we often find illuminating examples of the flaws within all of us that ultimately result in problems that could have been avoided had we stayed true to our original intent. For you Cliff Notes fans out there, this loosely translates to, “most of the time we do it to ourselves”. Aside from the life of a Hollywood starlet, is this any more true than in the data center design and construction process where the addition capabilities outside the project’s original scope wind up causing delays in the final delivery of the facility? No one maliciously intends on turning a six-month project into 18 months of added cost and finger pointing. In fact, these instances usually start with the seemingly benign phrase of “It’s just a…”
Perhaps only, “It’s not you it’s me” or “I just want to be friends” can wreck more havoc than any request that begins with the infamous, “It’s just a…” Of course we all know that it’s not the phrase itself, but it’s resultant dependent clause that leads to the problem. Typically, the initial request itself appears innocuous and even seems to make perfect sense—let’s add another generator, for example. And just like the witches’ prophecy to Macbeth—that he will become King of Scotland—most folks are pretty quick to jump on board. Unfortunately, this is where problems begin to creep in.
As Shakespeare so often demonstrates, a single action has many ramifications and it is no different within the realm of the data center. Adding new capabilities to the original data center design typically begins a cascading list of modifications or additions to support the original request. Are power requirements impacted? Do wiring schemes need to be modified? What are the modifications to the physical facility that need to be made to accommodate the desired “minor” deviation from the original site design? You see how these things can quickly add up. In Macbeth, his “it’s just a…” moment resulted in a fairly substantial body count. For you the metrics are cost and schedule impact.
In most instances requested changes, no matter how minor they seem, are the result of poor upfront planning. This is particularly an issue in “customized” data center implementations. While there are definitely instances (but not all) when an organization’s needs do necessitate a data center that is designed and built to address its unique circumstances, in the majority of instances these “unique requirements” are more the result of one of three things:
A) A person’s once in a lifetime experience that caused an outage 15 years ago (as if the world stayed in a vacuum state from that point)
B) Lack of actual experience, but someone told them they should have “it”
C) Hubris, because building stuff is fun and they really like career projects
than physical need. Like any major project, maintaining a discipline amongst all team members is the greatest hurdle to overcome. The best defense against “mission creep” or “saving us from ourselves”, take your pick, is a standard data center design and construction process for your application that is easily replicable and provides a consistent level of performance. For example, the standard, Facebook OpenCompute Design is not going to work for your 1MW hybrid mission critical data center.
As experience across industries has consistently demonstrated, a standard product is inherently superior to customized versions in terms of speed of delivery, reliability and long-term operation. This is due to the fact that all processes and components (including spare parts) are easily maintained and can be enhanced over time due to their ability to provide a regular feedback loop for analytical metrics. In contrast, movement along the “one off” continuum negates the ability of any of the elements to be effectively implemented. Thus, the near and longer term implications of even the most minor “it’s just a…” have a compound impact as they are adopted.
Designing and building a data center is obviously a non-trivial task, but comprehensive upfront planning and proven data center design can dramatically simplify the process. Modifications, even the “it’s just a…” type can cost you dearly in time and money. Maintaining the discipline to stick with your plan may be tough, but we can once again look to Macbeth as our cautionary guide. After all, no one wants to wind up with their head on Macduff’s pike.