How a single update to acquisition law can support cloud adoption
Government desperately needs a better way to buy cloud services. The Section 809 panel's call to create a new contract type for consumption-based solutions is a great start.
Twenty-five years ago, Congress passed the Federal Acquisition Streamlining Act of 1994, which created the notion of "commercial items" for government procurement. FASA was designed to circumvent the government's previous requirements to issue detailed specifications for items to be purchased, in favor of buying readily available commercial products and services. At the time, Vice President Al Gore famously went on David Letterman's show and dramatically destroyed an ashtray to call attention to the fact that, prior to FASA, a government contracting officer needed a multi-page detailed specification to delineate a compliant cigarette receptacle -- from which the government could neither deviate nor accept a substitute.
Frankly, the pre-commercial items era represented an absurdity bordering on madness. But while FASA has been great for the purchase of common commercial items like ashtrays, we have entered a new era of difficulty where the modern IT items the government wishes to purchase don't fit tidily into the commercial categories FASA and other regulations contemplate. At the core of the difficulty is that the government is forced to determine, prior to acquisition, if the thing to be purchased is a product or a service. That determination then drives the choice of contract type, the accompanying terms and conditions, and most importantly how the vendor is to be paid.
Unfortunately, the government's contractual dichotomy between products and services is inadequate for today's modern consumption-based solutions -- particularly the ones that are radically changing the face of information technology, like cloud. Is cloud a service, a product, or a hybrid of both? The inability to match the government contract type to the item to be purchased means the government cannot purchase cloud like an ordinary commercial consumer. This often puts the government in the awkward position of imposing many of the inefficiencies of old-fashioned computing on a solution that was expressly designed to eliminate them.
Agencies have few choices when it comes to paying for things in a dynamic "consumption-based" way. Think about this in ordinary consumer terms: Would it make sense to pre-pay a year's worth of funding to a service like Uber, in advance of actually taking a single trip? Often times this is how the government buys cloud services -- fully paid, upfront. And what if that upfront financial obligation locked you into a two-door Fiat 500 for all your travel, despite a potentially more capable all-terrain Land Rover becoming available at a lower cost? The government has little ability to take advantage of service changes or innovations that occur mid-contract, despite dynamic innovation being one of the most important value propositions of cloud.
Modern consumption-based solutions are not meant to be purchased in a rigid way. The solutions are dynamic and the contractual mechanisms to acquire them need to be equally dynamic. Currently, for the government, they are not.
Last summer, as a contractor supporting the General Services Administration's Cloud Center of Excellence program, I wrote about many of the inherent difficulties associated with contracting for cloud solutions. Fortunately, I haven't been the only one to notice the discord. In January, the Section 809 Panel, commissioned by Congress to investigate and create recommendations for the streamlining of DOD acquisition, put consumption-based contracting inefficiencies under the microscope and produced a quite useful recommendation.
In short, Volume 3 of the 809 Panel's report takes on many of the problems of IT procurement. "Recommendation 43" focuses directly on the issue of consumption-based solutions like cloud. The heart and soul of the recommendation is for Congress to create a new subcategory of services called "consumption-based solutions." These services would be defined as "any combination of hardware/equipment, software, and labor/services that together provide seamless capability that is metered and billed based on actual usage and predetermined pricing per resource unit."
Importantly, new services or features could be added to contracts at the discretion of the contracting officer, without necessitating a new competition. Bravo. This captures the essence of consumption-based solutions and allows for the government to legally accept dynamic innovation as it becomes available -- without having to anticipate or describe it in advance.
Perhaps most importantly, the panel recommends the creation a new contract type called a "Fixed-Price Resource Units," agreement to become the preferred acquisition method for all consumption-based solutions. This new vehicle type would establish base-line unit pricing, like an hour of computing, set an overall ceiling for the contract value, and then allow the government to pay on an incremental basis for consumption, in arrears, after actual usage. Brilliant. No more pre-payment. Goodbye to the problem of over- or under-committing funds for a solution, prior to actual usage. Frankly, this is the payment model that the National Institute for Standards and Technology has always required for cloud, but one that has been hard for the government to actuate.
The new contract type would also set maximum unit-pricing, but allow for the capture of spot-market price decreases, should commercial pressures force prices to fall. This fluidity of pricing, designed to take advantage of drops in cloud pricing, has the potential to produce immediate savings for the government. Equally important, the panel calls for Congress to provide flexible funding to allow for mid-year corrections, to avoid the problem of running out of money prior to the end of the year, and potentially breaching the Anti-Deficiency Act.
Kudos to the 809 panel. It may seem trivial to simply adopt a new category of item and create a new default contract type, but it's not. Fitting these new consumption-oriented solutions into extant government contracting models remains remarkably difficult. As a result, the government is inadvertently depriving itself of the highest value of these solutions, incurring unnecessary costs and missing the full benefit. Not only does recommendation 43 have the potential to radically accelerate cloud adoption and save the government money, it provides a framework to solve the contracting problems of future consumption-based solutions -- the ones we haven't even begun to imagine.