GovWin
 
 
Planning is the Key to Big Data Success

Through proper strategic planning and limited deployments to address narrowly defined business needs, federal agencies can gain valuable experience and develop competency with big data tools and solutions without taking on the financial risk associated with a major big data deployment.
In a segment that Federal News Radio aired at the end of September, Mike Parker, the Deputy Chief Information Officer for the Department of the Treasury was quoted as saying “The prescription [for successfully migrating a system to the cloud] truly is good planning, good business cases, and good governance.  That results in good outcomes." Some might view Mr. Parker’s statement as so obvious that it does not warrant noting. However, as anyone with experience working in federal IT knows, effective strategic planning can be a hit or miss proposition. The “hits” receive little attention. The misses show up in the press sooner or later. We all know these projects. They are the ones plagued by cost overruns, congressional scoldings, and public mea culpas from agency representatives.
Agencies are moving ahead with cloud computing adoption in various ways. Most of these are hidden from public view as work is either written into existing contracts or competed behind the veil of Government-wide Acquisition Contracts (GWACs) and multiple award IDIQ vehicles. Given the dearth of publicly released cloud computing strategies, we are left to assume (“hope” might be a better word) that the migrations are happening according to some plan or strategy. After all we have not yet heard of any colossal foul ups related to cloud adoption, right? My fingers remain crossed as far as the cloud transition is concerned. However, when it comes to the next big technology leap that agencies probably will need to make, this being the leap to big data tools and processing, agencies had better have plans in place or chances are that the transitions could be costly failures.
Why is planning how to use big data tools and processing so important? The answer lies in how “big data” is defined. Big data can be defined as any data set, structured or unstructured, that cannot properly be analyzed using traditional computing methods to glean insight that enables good business decisions. The key here is the phrase “traditional computing methods.” This phrase is a tip off that a new approach is required to properly employ big data tools and methods. Because traditional computing methods are inadequate, the implication is that agencies will need to spend money on new resources to handle massive and constantly growing data sets. These new resources potentially include everything from new dedicated hardware, network equipment to increase bandwidth for handling data, new advanced analytics software, and new professional specialists trained to analyze and report on the data.
The amount agencies will need to spend on big data will depend on the size of the deployment and, most importantly, on the business need that the agency customer has defined. Business needs translate into business cases and business cases require planning. The use of big data analytics by the Centers for Medicare and Medicaid Service (CMS) for the purpose of reducing the agency's $50 million in annual improper Medicare payments is a case in point. CMS is using a big data tool to solve a specific business need that it has. This kind of limited deployment gives the agency experience with big data tools and methods that it can build on. It would be correct in this context to argue that the implementation of big data solutions is most successful if the growth of the solution is evolutionary as opposed to revolutionary. By taking an evolutionary approach, agencies gain valuable experience and develop competency without taking on the financial risk associated with a major big data deployment.
Consider in contrast the potential expense of taking a “revolutionary” approach to big data. The computing power required for some scientific purposes boggles the mind. Acquiring this computing power would require either buying a supercomputer or scaling up a very large number of servers. Then there is the need to store all of the data and to push it through the network. Most networks presently cannot handle the bandwidth required. The Energy Sciences Network (ESnet) at the Department of Energy is a good illustration. As Daniel A. Hitchcock, Associate Director of the Office of Scientific Computing Research at the DOE recently told a conference audience, in order to facilitate data sharing and analysis between scientists, the DOE has had to develop a 100GB per second network. Yes, you read that correctly, 100GB per second. Work on ESnet has been underway for many years now, so it has the benefit of being a legacy network built when funding was available. Today, however, what agency has the money to build something like ESnet?
If we stop for a moment and consider the potential cost of not taking baby steps toward big data then the wisdom of strategic planning becomes clear. In today’s budgetary climate federal agencies simply do not have the luxury of taking a revolutionary approach to any technology, big data or otherwise. 

 

Comments (Comment Moderation is enabled. Your comment will not appear until approved.)