In the ongoing effort to consolidate the government’s thousands of data centers to a more manageable number, the Defense Department may have offered the coup de grace.
In a July memo, the department’s chief information officer declared that all core functions — including storage, computing, cloud services — are to be delivered via the Defense Information Systems Agency’s Defense Enterprise Computing Centers.
The announcement, in effect, simplifies consolidation for the military services, giving them a predefined destination for their future data center operations. Still, the military branches have a lot of heavy lifting to do outside the DISA DECCS, and they face diverse challenges in their efforts.
In the big picture, the Army says data center consolidation is on track, despite some initial questions about definitions.
Early on, the service had 185 centers, of which 130 have so far been closed. A complication arose, however, when the Office of Management and Budget redefined “data center” more broadly, boosting the Army’s inventory to 800 centers.
Despite the shift, a self-imposed target of a 60 percent reduction by the close of fiscal 2018 remains on course, said Neal Shelley, the Army Data Center Consolidation Plan lead.
“We have met quota in FY 2011 and 2012, and we are at 95 percent of quota for 2013,” he said. “We will beat our goal of 66 closures in FY 2013.”
As the DISA DECCs directive unfolds, it will help give direction and definition to the efforts, but it won’t cut the Gordian knot, said Jeremy Hiers, Army’s project director for enterprise services.
“What is the scope of DISA in this? There are a couple of items in the policy that leave room for exceptions,” he said.
One key question: What computing services will remain outside the DISA framework? The Army and DoD are exploring this together.
Another issue is obsolescence. How many systems are so far along in their life cycle as to make them not worth moving?
Again, the issue is under analysis.
Perhaps most significantly, Army officials may feel compelled to bypass DISA when consolidating services could compromise field capabilities.
“In our business, light-speed matters,” Shelley said. “There are things you cannot do remotely. The response time is not sufficient or it is so critical that you cannot have that application running several miles distant. So there will continue to be data processing going on at posts, camps and stations.”
This question of proximity could be a major sticking point as the services try to sort out their most likely consolidation scenarios. That future infrastructure will be based largely in the cloud, and that may give pause to commanders in the field who worry about operational integrity.
“When you have a war fighter who needs a mapping data service, for example, being able to deliver that over a cloud service becomes a completely different situation” versus having a local server, said Gunnar Hellekson, chief technology strategist of the public sector division at open-source developer Red Hat. “You have to have a large amount of bandwidth, you have to have redundancy, while on the operational end you may have someone with just an intermittent satellite connection.”
As consolidation continues, military information technology planners will need to strike a balance between the need for more efficient structures, and the operational imperative, said Hellekson, whose company plays a role in nearly every major DoD IT program.
Hand in hand with this come concerns about the changing nature of those situations.
“If things were static, it would be a heck of a lot easier,” said Kevin Kelly, CEO of contractor LGS Innovations.
“Where is the military going to be 24 months from now with its deployed forces? It’s difficult to say, and you certainly can’t just place a data center in every corner of the world.”
While issues such as mobility could be a deciding factor in the where and how of consolidation, outside contractors involved in the consolidation effort say the military branches could face other hurdles.
Territoriality may well be a sticking point, said John Garing, a vice president at ViON Corp. As DISA CIO at the turn of the millennium, he led a consolidation to DECCs that resulted in a reduction of 2,500 people and $300 million in annual operating costs. He predicts the military will meet some resistance from field officers personally attached to their servers.
“There are pockets of box huggers out there,” he said.
Installations with highly specific missions, for instance, may be especially reluctant to toss their server functions into the common pot.
The Army acknowledges the hazard. With each command managing its own data centers, every server “is often considered a jewel,” Shelley said. “It was built because it fulfilled a need, and naturally people think that if they are in control of something, they are the ones fulfilling that need. So we are changing lifestyles, changing cultures, asking people to rely on someone else to deliver services that they have always considered to be mission-critical.”
Considering the variables — DISA directives, localization, territoriality — it becomes increasingly difficult to answer a core question: How’s it going? While the Army, for instance, claims steady forward momentum, others still ask what exactly constitutes progress when it comes to data center consolidation.
“Throwing out numbers can be misleading,” Hellekson said. “If a data center is a single server in a closet, it doesn’t make sense to talk about losing 100 data centers here or 300 there. It is not usually apples to apples, so it’s difficult to say which agencies are doing ‘better’ than others.”
Data center consolidation “is not an exercise where you say you are 50 percent or 60 percent finished,” he said. “It’s a process where, over time, you get better at managing these systems.”
Moreover, consolidation is a moving target. “The challenge with IT is that it changes so much and so often,” said Angie Heise, vice president of enterprise IT solutions at Lockheed Martin, which is a key player in a $250 million Army cloud contract.
“The way all of these things look and operate today, this is not necessarily the way we need them to look and operate 10 years from now,” she said. “It makes it hard to say that you can put a bow on it at some point in time.”