US Government: Playing Jeopardy! with the Cloud

One of the most popular features of our Licensed ZapThink Architect (LZA) SOA course is the Jeopardy! game at the end. With clues taken from the material, the responses as per the television show must be in the form of a question. But while answer-first, question-last pattern may be amusing for a game show, it’s a serious pitfall in the world of IT. In fact, understanding what the proper question should be is an important lesson in the LZA course, as it’s a critical skill for any architect.

All too often, however, organizations start with an answer while being unclear on what the question is. Case in point: the US government’s Cloud First initiative. Originally part of former Federal CIO Vivek Kundra’s 25-Point Plan to Reform Federal IT Management, Cloud First called for government agencies to move three existing applications to the Cloud and consider Cloud for any new IT initiatives. Why they would want to take such actions, other than simply to comply with the initiative, wasn’t clear.

In other words, the answer was Cloud First. But what was the question?

Problems with a Missing Business Case

Pushing agencies to rush apps to the Cloud without a clear business case led to a scramble to find the “low hanging fruit” across the government. According to Federal CIO Steven VanRoekel, “You’d see people taking a small solution, and moving that to the cloud and saying ‘we’ve done our piece, we’ve achieved our goal.’” In other words, agencies would try to find the simplest, cheapest and least disruptive way to comply with the Cloud First mandate, instead of focusing on solving a real problem. Furthermore, the Government Accountability Office (GAO) only tracked the initial push to the Cloud, which mostly consisted of simpler applications like email and public-facing Web sites. No agency felt that it was practical to deal with more difficult challenges, namely migrating complex legacy systems, and nobody was holding the agencies to account.

Even with these low-hanging fruit, furthermore, the business case was often unclear. Dr. Paul Tibbits, the Deputy CIO for Architecture, Strategy and Design at the Department of Veterans Affairs (VA), recently pointed out this problem. “We have got to show how we are going to get more value for the money we spend,” Tibbits said. “It is not 100 percent clear that expenses go down if we jump into the Cloud. The revenue stream is up there in neon lights, we have got to figure out if that is going to save us money or not.” Even the conventional wisdom that the reason to move to the Cloud was to save money wasn’t always holding water.

Ironically, the VA recently backtracked on moving their email to the Cloud. Earlier this month, the VA terminated its five-year, $36 million cloud computing contract with HP Enterprise Services for Cloud-based email, citing a material change in the agency’s requirements. At this point in time it’s unclear what requirements had changed, but it’s unlikely that the VA’s functional requirements for their email were the problem. Far more likely is that even with the cancellation penalty taken into account, the VA realized that they would actually save money by keeping their email in house vs. moving it to the Cloud.

To make matters worse, it has been difficult to gauge the success of Cloud First, because the federal government hasn’t been tracking key indicators, in particular, the number of applications they’ve moved to the Cloud or the amount of savings they’ve generated. Then again, which is more important: the number of apps or the savings? Clearly, simply counting the apps an agency has shoehorned into the Cloud can mask the fact that they are moving the wrong apps simply to check them off the list, without a clear view of the business case – a problem the government has come to realize. According to Scott Renda, Cloud Computing portfolio manager at the Office of Management and Budget (OMB), the OMB will be “placing less emphasis on counting services and more emphasis on business value. I don’t anticipate, at this point, coming out with a new target about ‘move X more services to the Cloud.’”

Pinning Down the Business Case

Saving money, of course, is the primary reason the government touts for moving the Cloud, and there have been several success stories, in spite of the VA’s U-turn. In fact, Federal CIO VanRoekel recently said he’s worked with agency CIOs to shave at least $300 million in IT costs, through Cloud, data center consolidation, and elimination of redundancy – an amount that pales in comparison to the federal government’s $79 billion IT budget, but a good start nevertheless.

On the other hand, just a few days ago, Dave McClure of the General Services Administration (GSA), David Powner of the GAO, and VanRoekel faced a Congressional hearing, unable to even identify how many non-core data centers the government owned. Was it 3,000? Or perhaps as many as 7,100? Depends on what you mean by “data center,” according to VanRoekel. Furthermore, the government owns 580 redundant financial systems and 777 redundant supply chain systems, together costing about $6 billion – presenting a clear business case for a legacy migration strategy that would indubitably include a move to the Cloud over a multiple-year timeframe.

In spite of these challenges, many agencies have already shown real cost savings. Cloud successes include the US Department of Agriculture (USDA) Food & Nutrition Service Supplemental Nutrition Assistance Program (SNAP), formerly known as Food Stamps. The SNAP Retailer Locator has been in the Amazon cloud since 2010, costing the federal government a few thousand dollars in hosting fees per month, according to Jonathan Alboum, CIO of the Food and Nutrition Service.

The Department of Health and Human Services (HHS) has perhaps the best Cloud track record across the government so far. According to a GAO report, by July of 2012, HHS  was one of only seven agencies to have successfully incorporated Cloud Computing requirements into their policies and processes at that time. The apps they had successfully migrated include grants solutions site GrantSolutions.gov, Administration for Children and Families audit resolution tracking, MedWatch Plus, HealthData.gov, and more than 275 apps running on the FDA Private Cloud.

Interestingly, the business case for moving these applications to the Cloud isn’t simply a matter of cost savings. HHS CIO Frank Baitman cites elasticity as one of the primary benefits HHS has achieved by migrating these apps: the ability to deal with unpredictable spikes in demand without having to pay for underutilized servers. Such a business case requires a greater focus not only on the desired business value, but on tracking the realized value over time. It comes as no surprise, therefore, that one of HHS’s secrets to success was the fact that they tracked estimated costs, major milestones, and performance goals for their Cloud-based apps.

While the HHS’s Cloud efforts have focused on IaaS, SaaS is also a large part of the government’s Cloud strategy. For example, the National Archives, the GSA, the Department of Energy’s Idaho National Laboratory and the National Oceanic and Atmospheric Administration (NOAA) have all moved to Google Apps for Government. Google Apps for Government is essentially a government-focused community Cloud that offers the same functionality as Google Apps for Business. The GSA for one expects to save 50% of their operating costs every year with their Cloud efforts.

Placing Cloud into Context

Perhaps the most important lesson from shifting the focus from the answer (Cloud First) to the question (the business case) is that for a given business problem, Cloud may or may not be the right answer. After all, there are many other approaches for addressing IT problems than simply moving to the Cloud.

The new CIO.gov Web site highlights this increasing focus on addressing key business issues rather than simply promoting the buzzword du jour. The focus of the federal CIO’s office, at least according to this site, is on maximizing value, which for the government means data center consolidation, shared services, and greater visibility into the workings of IT in the government in order to identify areas of improvement. Cloud Computing may support any of these efforts, true, but the discussion rightly begins with identifying problem areas.

In fact, this month the government announced savings of $885 million over the last year by consolidating redundant systems. The key enabler of this savings is an approach the government calls PortfolioStat, where officials review spending data on IT resources with an eye for duplication. PortfolioStat is helping the government avoid the embarrassing problem of not knowing where all their systems – or even their data centers – are located. The Social Security Administration, the Department of Defense, and the Department of Homeland Security have each reported savings in the hundreds of millions of dollars through PortfolioStat.

TechStat, a similar program to PortfolioStat that focuses on the performance of troubled IT projects, has indicated a $4 billion savings over 2010 and 2011. However, the GAO questions the validity of this number, as they found no way to validate the data, and only a small portion of troubled IT projects received a TechStat review. Nevertheless, both PortfolioStat and TechStat focus on identifying problems, where Cloud First centers on one of many possible solutions. PortfolioStat and TechStat aren’t nearly as sexy as the Cloud, but they’re essential elements of an effective plan to bring down IT costs and get more value out of the IT the government owns. And as taxpayers, we should all be happy about that.

The ZapThink Take

There is an important lesson here for any organization, public or private sector, who wishes to move to the Cloud. Remember, you don’t actually want to move to the Cloud at all. You want to solve your business problems. If you’re spending too much on IT, or you have redundant systems, or perhaps your problem consists of unpredictable spikes in demand, then yes, the Cloud may be the right answer. But don’t play Jeopardy! with your IT decision making processes. Start with the problem and identify the appropriate solution. Never start with the answer and try to figure out the question.