Why data centers are hard to count

In response to a July 24 FCW article highlighting the severe underestimate of the number of federal data centers, a reader who identifies himself as Peter Marshall writes:

It has been my experience in the field that not only each federal agency, but offices within each agency, and even down to the Division level have different definitions for the "Data Centers" proper. Many will define small data processing and IT equipment rooms ranging from 500 square feet to 3,000 square feet as data centers when the dynamics of the facility's primary systems and spaces allocated to data processing do not have the characteristics of a "Data Center" as the industry defines it. Therefore, OMB in conjunction with all Federal Agencies and IT components need to first gather consensus around definitions and categories of data processing spaces, rooms, and facilities prior to the development of a fed-wide inventory of "Data Centers". The biggest challenge is getting past senior management who think they are IT savvy and the IT authority within their sphere of control, when in fact their technical experience and understanding of data processing systems, facilities and strategies are limited. It is my guess that once you define the Data Center you will find that the actual numbers of Data Centers (proper) will be down and the number of other IT and data processing spaces/rooms will be up. But until then there will continue to be a disconnect between the truly IT savvy technical engineer and the upper level manager.

Frank Konkel responds: Your criticism is spot on, Peter. When pressed at the data center hearing before Congress in July about how the Office of Management and Budget underestimated the number of data centers so dramatically, the federal CIO said the dramatic increase in data center count was in large part a result of a change in the definition of the term. Previously, OMB measured data centers based on size – greater than 500 square feet – but that definition was changed to include "data centers" that might fit in your coat closet, as you allude to.

The loudest criticisms of OMB regarding data center consolidation in recent months have been the agency's lack of leadership and lack of guidance in metrics to track cost savings realized. The criticism keeps coming from Congress and from reports generated by the Government Accountability Office, which publicized the data center increase before OMB owned up to it despite reports that the agency knew about the higher data center number  as early as mid-2012. And OMB's efforts at mitigating the criticism so far have fallen short. Now that FDCCI has been rolled into PortfolioStat, another IT initiative led by OMB, the agency is intent on "optimizing" rather than closing data centers, yet OMB hasn't produced any metrics that agencies could use to track savings, leading to minimal results for an initiative that launched in 2010 and was supposed to save between $3 billion and $5 billion by 2015.

As for the disconnect between IT-savvy engineers and their managers, I agree, but I think the biggest communication barrier is among OMB, GSA and the agencies charged with carrying out these executive orders and OMB mandates. If the agencies aren't given metrics to follow and repeatable plans officials can alter to fit their agency, what is the point of the initiative in the first place? Because it isn't going to save a lot of money if agencies have to figure it all out on their own.