US Military Asks for Private Sector’s Help to Understand Cloud Computing

In spite of having developed the base for cloud computing – the Internet – the US military seems to be lagging behind the civilian sector in this regard. And this is especially galling for Defense Advanced Research Projects Agency (DARPA), the US Department of Defense’s advanced research arm that spearheaded the technological innovations mentioned earlier. Now, in an effort to remedy the situation, DARPA has reached out to the private sector for assistance by issuing a formal Request for Information (RFI).

It’s not that the US military has been completely silent as regards to cloud computing. There have been intermittent efforts to understand and co-opt this new technology (See: NATO set to rule the cloud with IBM’s help and What Does the US Military Want from Cloud Computing? ). However, with this RFI, the military is taking a bold step towards admitting its shortcomings and trying to rectify them, and it’s not above asking others for help.

The RFI’s description outlines the scope of the exercise: “In the civilian domain, cloud computing has demonstrated substantial benefits in development and maintenance of software applications. To date, the cloud computing paradigm has not been effectively exploited in embedded military applications, for reasons related to performance and correctness constraints. The Defense Advanced Research Projects Agency (DARPA) Microsystems Technology Office (MTO) is requesting information on techniques through which the benefits of cloud computing can be obtained for embedded military systems”.

While cloud computing companies are familiar with what “cloud computing” stands for, the document defines “embedded military systems” as “computing hardware and software that is part of a larger system with a specific military purpose”. It enumerates the benefits of using cloud computing in embedded military applications as:

• Cost efficiency: Recurring and non-recurring costs may be decreased.
• Manufacturing agility: The time to design, implement, and deploy embedded systems may be reduced.
• Reliability: Use of pooled resources may remove single points of failure.
• Maintainability: System maintenance costs may be reduced since fewer unique systems are maintained. Upgrading both hardware and software may be facilitated by the reduced dependencies among system components.
• Democratization: Since arcane optimization skills are less useful on cloud platforms, more people may be able to program for them.

At the same time, the document shows that the US military is fully cognizant of the challenges in such efforts, listing them as:

• Efficiency: Software that is designed and written without knowledge of the specific computing platform on which it will execute is generally less optimized than platform-specific software.
• Security: Primarily, information must not be allowed to propagate where it was not intended.
• Reliability: Embedded applications are often subject to stringent reliability and availability requirements.
• Performance: A salient performance requirement in many embedded systems is real-time response to external stimuli.
• Verifiability: Performance according to specifications must be guaranteed with high degrees of certainty. Such guarantees are required for initial deployment and after upgrades.

The RFI then outlines DARPA’s expectations from the exercise, “information on innovative technologies that will overcome the preceding obstacles and obtain the benefits of cloud computing for embedded military applications…..solutions that are revolutionary, with high payoff and potentially high risk, and can be deployed into practice after 3-5 years of intensive research and development”. It also clearly mentions what DARPA does not want – “evolutionary improvements to existing approaches, methods, or practices”. The agency mentions the possibility of an invitation-only Industry Day to discuss the topic further.

It is clear that DARPA has thought long and hard on this subject, given its enumeration of advantages and disadvantages. However, what’s even more interesting is that it recognizes the wide implications of cloud computing and makes its preference for revolutionary over evolutionary changes quite clear.