It's time to burst the cloud's hype bubble
Agencies are rushing into an unfinished technology
Michael Daconta (firstname.lastname@example.org) is the chief technology officer of Accelerated Information Management and the former metadata program manager for the Homeland Security Department. His latest book is, “Information as Product: How to Deliver the Right Information to the Right Person at the Right Time.”
As cloud computing has become the silver bullet du jour, we suddenly are faced with two simultaneous and equally dangerous phenomena: “Everything is Cloud” and its evil twin, “Cloud is Everything.”
The solution to both of these phenomena is to take a lesson from the old Rolling Stones song and kick some initiatives and concepts off of the cloud.
First, let’s address the notion that Everything is Cloud, or, as a recent GCN article put it, “agencies are becoming more cloudlike.” I recently read an article on the “Cancer Knowledge Cloud” sponsored by the National Cancer Institute. It is a grid computing initiative called caBIG, rebranded as a cloud offering.
Don’t look down: The path to cloud computing is still missing a few steps
I first heard of caBIG in 2005 when I was at the Homeland Security Department. The problem is that the standards, interfaces and specifications for grid computing, developed during the past five-plus years, are different from the set of standards, interfaces and specifications for cloud computing. So in reality, grid computing is not cloud computing unless you squint really hard.
Try comparing the caBig and caGrid application programming interfaces to Amazon Web Services offerings, and you will find it difficult to know where to begin. Repeat the exercise with Windows Azure or Google AppEngine, and you will likely have the same results. I see this Everything is Cloud or cloudlike response from agencies as a natural reaction to misguided pressure to prematurely leap to unfinished technology. After all, if you ask a stupid question…
Now, let’s examine the other side of the coin. I recently viewed, with slack-jawed amazement, an Army Leader Training video in which the cloud saved the day with instantaneous sharing, app store simplicity and precisely relevant data raining down from the cloud like manna from heaven.
Such naiveté is just downright dangerous. It is high time to burst this hype bubble with a cold splash of reality, which will reside in those devilish details.
For the federal government, much of this difficult, painstaking work has been handed off to the National Institute of Standards and Technology, which recently announced the Standards Acceleration to Jumpstart Adoption of Cloud Computing initiative. In a nutshell, NIST will set up a Web portal, post existing specifications, create or encourage reference implementations, and post gaps in specifications that the community needs to fill with new specifications. All those specifications are then thrown over the wall to standards organizations.
Although many parts of this sketchy plan sound good, it smells like a Web 2.0, market-driven approach to standardization that is too Pollyanna-like to actually work. It’s missing a governance process that can winnow the proliferation of cloud requirements and specifications down to only the best and most fitting for government use.
First, winnowing down the requirements will tamp down the ability to state that the Cloud is Everything. No, it is not — there is a properly scoped subset of computing functionality that is well suited for the elasticity of a cloud environment. Additionally, the privacy and security constraints for government cloud computing are much different than those for commercial cloud computing, and those differing requirements must result in different technical specifications.
Second, winnowing down the technical specifications will tamp down an agency’s ability to rebrand existing initiatives as cloud computing. Why is that important? Because the huge benefits in cloud computing require a cohesive stack of technologies that must support and reinforce one another in a coordinated manner. Such a single stack cannot be built by bandaging existing pieces together for reasons of expediency.
Winnowing down possibilities and selecting one interface or capability over another are hard work. It requires a deep technical understanding of the issues to decipher which solution is best. Bruce Lee said it best, “In building a statue, a sculptor doesn't keep adding clay to his subject. Actually, he keeps chiseling away at the inessentials until the truth of its creation is revealed without obstructions.”