Progress in Federal Data Center Consolidation

September 6, 2012 3 Comments »
Progress in Federal Data Center Consolidation

Since the inception of the Federal Data Center Consolidation Initiative (FDCCI) in early 2010 under then Federal CIO Vivek Kundra, The Data Center Journal has covered the progress of this initiative with a due amount of skepticism. Although some progress has been made—likely despite resistance at the agency level—the project still faces fundamental barriers to success.

Steps Toward Fewer Data Centers

An initial tally of federal data centers stood at around 1,100 in 2009 according to a federal CIO memo; that tally later increased to around 2,100 by late 2010. The current count is over 3,000 data centers, where a data center is currently defined as “a closet, room, floor or building for the storage, management, [sic] and dissemination of data and information.” This definition, unlike the previous one, no longer considers area or tier classification. According to DatacenterDynamics (“US government launches data center services marketplace”), the current goal of the FDCCI program is to close about 1,200 data centers, with about 320 closures having been achieved thus far—slightly over 25% of the goal. About 680 closures total are slated for completion before 2014.

The latest step in the quest to reduce the federal data center footprint is a marketplace designed to enable agencies to buy and sell data center rack space, with the eventual addition of virtualized server space. DatacenterDynamics notes, however, that “the main obstacle to [the marketplace’s] further development is implementing a system of enforceable inter-agency service level agreements (SLAs) and metering.” Part of the problem is simply the culture of government agencies: profitability is not a concern, so these agencies are not motivated to deal with other agencies as customers in a mutually beneficial arrangement. “Federal regulations prohibit a provider agency to make a profit,” the article further points out.

The lack of enforceable SLAs is particularly troublesome for agencies, such as the Department of Defense, with mission-critical applications. Understandably, such agencies would have some reluctance to rely on other agencies with less concern for meeting certain minimal service standards. The difficulty is implementing some system that provides incentives for agencies; since profit is essentially out of the question, the remaining option seems to be on the negative side, such as penalties for failing to meet some goal. Here, human nature becomes a monkey wrench: a system that only involves punishment for failure, but no reward for success, is more likely to produce resistance rather than progress. And, overall, government agencies generally have no positive incentives for performance anyway, since their clientele is almost always a captive one. (Would anyone argue that UPS or FedEx could do better than the U.S. Post Office in delivering mail, were the Post Office not granted a legalized monopoly?)

A Fundamental Problem for FDCCI

The FDCCI has seen some ostensible progress toward its goal of a reduced data center footprint, and more progress seems to be in the works. This is not necessarily a great feat, as the federal government is inefficient to the extreme, but it does represent at least token movement in the right direction. But the problem that the FDCCI faces is more fundamental.

Imagine a growing company with ever-expanding IT needs. Ignoring for the moment the option of outsourcing via the cloud, for example, imagine that the company wanted to reduce its data center footprint. If you assume that it’s not overly inefficient in its current footprint, its goal would appear rather incongruous: a growing company needs more facilities to deliver services (whether internally or to customers), not fewer. In the same vein, a growing government cannot hope to truly reduce its data center footprint when demand for IT services is on the rise. The FDCCI may not be concerned with this, however; nevertheless, barring a reversal of the trend of ever-growing government, demand for IT resources will eventually reverse reductions in data center footprint. An alternative, of course, is outsourcing through the cloud—but this isn’t really a reduction in the government’s data center footprint.

Furthermore, data center consolidation isn’t free. One of the often-cited reasons for the project is to reduce federal spending on IT, but as notes (“Congress Urged To Hold Hearing on Data Center Consolidation”), a Congressional Research Service document raises the concern that “somebody has to watch the up-front costs of data center consolidation, because it does in fact require significant investment. In fact, in some cases, the costs may outweigh the future savings.” Numbers quantifying savings from this or that effort in federal IT are regularly trotted out, but such data may be akin to claims by the Obama administration of “creating” 4.5 million jobs; such numbers don’t take into account losses. Thus, the question for the FDCCI isn’t money saved, it’s net money saved.

Why Pick on IT?

You may have heard the little joke about fast-food patrons ordering a burger, supersize order of french fries and a diet soft drink. In some ways, the situation with IT in the federal government is similar. This isn’t to say that data center consolidation and increased efficiency is a bad or even unimportant pursuit, but it must be kept in perspective. Just as bad eating isn’t balanced by a less bad beverage, so projects like the FDCCI will have no substantial effect on the budget of the federal government. The largest budget items are currently Medicare/Medicaid and Social Security. To be sure, these budget items include their fair share of IT needs, but the point is that it’s not IT per se, but entitlements that are breaking the bank. With a national debt having topped $16 trillion, and no end in sight for trillion-dollar annual deficits, a billion-dollar cut here or there in IT is hardly worth noting. A real solution to burgeoning federal IT consumption is major, extensive, permanent and (likely) painful cuts to entitlements—and to defense spending. When the government has less to do, it will have less need for IT.

In the meantime, again, consolidating existing IT resources to the extent possible is not necessarily unwarranted, although the savings to be gained from the effort may well fall far short of expectations owing to implementation costs. The recently launched marketplace for data center rack space makes some sense—at least conceptually. The culture of government agencies, which is completely devoid of any drive to profitability or even interagency cooperation, remains a stumbling block both for the new marketplace as well as any attempt to implement real reductions in consumption.

As the FDCCI approaches its goal, one question to investigate is whether the data center closures are substantial. Recall that the FDCCI definition of a data center changed recently; although this increased the tally of data centers, it also likely included many inconsequential computer rooms and closets that many IT professionals would hesitate to denote as data centers. If closures largely consist of these instances, then even a large number of such closures may not add up to much real reduction in data center footprint. Thus, the news on the FDCCI front is somewhat mixed: the federal government is making headway, but the tangible benefits of that progress remain in some doubt. Implementation costs and the extent to which the reported data center closures are actually substantial may weigh significantly on any gains from the project.

Photo courtesy of C.E. Kent

About Jeff Clark

Jeff Clark is editor for the Data Center Journal. He holds a bachelor’s degree in physics from the University of Richmond, as well as master’s and doctorate degrees in electrical engineering from Virginia Tech. An author and aspiring renaissance man, his interests range from quantum mechanics and processor technology to drawing and philosophy.


Leave a Reply

Pin It on Pinterest

Share This