Alternatively, were there questionable requirements in the NSA pro forma that the designers would have liked to change, but weren’t allowed to? Were such decisions based on cost, politics, or lack of knowledge within the Corp of Engineers that managed this project?

Either way, if the professionals were pulled away from their goal, or prevented from using their expertise to the fullest, then the present problems may be only the tip of a giant iceberg that could end up eventually sinking a titanic project. Let’s hope not.  But when these things happen, designers, developers and contractors have to speak up or bear the consequences of remaining silent. Let’s hope that if there were such occurrences—and it’s hard to believe there wouldn’t be on a project like this—that the designers and contractors prepared good lifeboats for themselves in the form of disclaimer letters.

Q 3:     Was the schedule simply too aggressive?

There shouldn’t be much speculation about this. It really goes hand-in-hand with Question #1, but seems worth of addressing separately. Every announcement and report has noted the aggressive schedule. The contract was awarded on Sept. 24, 2010.  Groundbreaking was only four later on Jan. 6, 2011, and opening was supposed to be Sept. 24, 2013, just three years after contract award. That’s not very long for a million square foot government project of this complexity, even under a “Design/Build” contract where construction begins before designs are completed—a dubious practice in complex mission critical work anyway.

Private enterprise might be able to accomplish this by hiring the very best, paying them appropriately, and leaving them unconstrained. Google has certainly done it. But government doesn’t tend to work that way—at least not until the problems need to be fixed, as we are also seeing with the healthcare website.

Did the bidders erroneously assume that, as with so many government contracts, the time frame would be extended after award of contract? It certainly has been now, but not because the government recognized the pitfalls of over-pushing the schedule on a project of this importance and magnitude.

The old adage applies:  “You can have it good, fast, cheap.  Which two do you want?” The NSA clearly wanted all three. It’s axiomatic that this data center has to be very good, but the NSA also wanted it fast. And the government always wants it from the lowest bidder, which supposedly protects the taxpayers but somehow always seems to cost us more. Did the bidders just ignore the realities of what could actually be accomplished in the face of a prestigious contract?

Q 4:     Have we reached a realistic limit in data center size?  Is this thing simply too big to work?

We should be able to at least partly answer this one. Plans that have been published show four data halls, each apparently with its own utility plant. That’s how most new, large-scale data centers should be, and are being designed and built. And since quite a number have been successfully completed this way, we know it can be done.  So the answer to this question should be that it is certainly not too big to work. This “big” data center should actually be four separate smaller facilities that happen to be housed in a common building complex.

So is there a point at which all four of these otherwise “independent” data centers rely on a common infrastructure?  At 65MW they probably shouldn’t—at least not subsequent to the main power grid, which should have at least two substations. But then, if there’s no common infrastructure, why are these reported “power faults” stopping everything? What’s preventing the NSA from moving into the sections that aren’t affected? Was a shortcut taken that subjects all four entities to the same disruption? Or is there an endemic problem with the design, or with some piece of marginal equipment that has been replicated in all four segments? If so, shouldn’t it be relatively easy to find the cause?