How do you overlook 3,000 datacenters?

It takes a major bureaucracy to lose track of where their data processing reside
Written by David Chernicoff, Contributor on

While the attention has been focused on the new NSA datacenter in Utah, a re-evaluation, 3 years into the Federal Datacenter closure program, identifies an additional 3,000 facilities that fall under the closure consolidation mandate. Granted, under the loose definition of datacenter that the government is using that could mean 3,000 racks hidden in 3,000 utility closets throughout the country, but how can any organization not know where its data processing and storage facilities actually reside?

As reported in the Federal Times, David Powner, Director of Information Technology Management at the Government Accountability Office, told Congress that a recent estimate of the number of datacenters subject to closure had reached 6,000, which included 3,000 that had not previously been counted.  He followed that up by saying that after three years there were still no good, hard numbers on the total number of datacenters in use.

This kind of information really highlights the difference between IT operations in government and private business.  While there are stories of occasional servers being misplaced or even walled up in the business world, losing track of entire datacenters, regardless of how small, is something that simply wouldn’t happen in a business environment. With multiple agencies running their own IT and no explicit oversight that gives authority or responsibility to anyone further up the government chain there is rarely anyone with authority who can be held responsible for this poorly organized government IT effort.

With the original goal of shutting down 800 of the reported 2,100 datacenters that the government thought they were running three years ago becoming a less than tangible goal, the OMB has now decided that the target is 40% of the identified datacenters. The OMB is also now specifying that only “non-core” datacenters, be shut down, which could be a useful metric if they had bothered to define the term “non-core” so that it could be applied equally across the government departments.

For IT personnel who have long dealt with mandates from on high that rarely took the reality of the situation into consideration, these issues are unlikely to be much of a surprise. With a mandate handed down from the OMB without details or significant investigation into what implementation would actually require, the odds of this program hitting its completion date of 2015 seem to be very small.

Editorial standards