An analysis of behavioral economics applied to cloud computing adoption triggered an enthusiastic response from some cloud enthusiasts. However, it might be prudent to remember that the analysis applies equally to adopting the cloud...and leaving it.
James Urquhart is a field technologist with almost 20 years of experience in distributed-systems development and deployment, focusing on service-oriented architectures, cloud computing, and virtualization. James is a market strategist for cloud computing at Cisco Systems and an adviser to EnStratus, though the opinions expressed here are strictly his own. He is a member of the CNET Blog Network and is not an employee of CNET.
Understanding the economics of cloud computing is critical to driving the right operations decisions for IT organizations of all sizes.
Some months ago Joe Weinman, vice president of corporate strategy at AT&T, posted a primer of sorts explaining the mathematics of computing utilities. When should an IT organization choose a public cloud computing or storage utility? Should you select dedicated systems for your application, or perhaps a hybrid cloud envionment, combining public clouds with internal cloud computing resources? Weinman's post has formulas that can help answer these questions and more.
Last week, Weinman followed up that post with another look at cloudonomics, but this time from the perspective of behavioral economics, relying heavily on the work of Dan Ariely, a professor of psychology and behavioral economics at Duke University. Ariely wrote a book called "Predictably Irrational," which investigated a broad range of irrational biases in human behavior.
Weinman took the most relevant of these biases and applied them to decisions about adopting cloud computing. As you might expect, there are a variety of interesting conflicts between what is economically preferable and emotionally comfortable. For example:
The Endowment Effect -- People value goods that they already own more than they would pay to acquire them. Ariely showed that for the same hard-to-acquire Duke basketball tickets, students were willing to pay up to about $170, but weren't willing to sell them for less than $2,400. Add in the choice-supportive bias, which rationalizes selected options and discounts unselected ones, and a stubborn fondness for existing IT technology and organization assets can be understood.
As this example shows, there are plenty of excuses IT managers can find for not adopting cloud, despite the overwhelming evidence that cloud is the better economic choice much (if not most) of the time. Weinman's post outlines 10 such "cognitive biases."
Another respected cloud blogger, Hyperstratus CEO Bernard Golden, added to the discussion with an impassioned plea to IT managers to recognize these biases and embrace cloud wherever it makes sense:
Certainly, the published case studies and examples would seem to dictate that cloud computing should be aggressively considered as the foundation of infrastructure initiatives; however, all too often one encounters an attitude that can only be characterized in a fashion akin to a joke ascribed to economist reactions to various real-world outcomes: "Sure it works in reality, but will it work in theory?"
I tend to agree with both Weinman and Golden, but there is more to this subject than the decision to move public clouds. In fact, while making a case for why public cloud adoption would be easier sans human nature, they both leave the impression that the only irrationality is around embracing that deployment model. The truth is that those adopting public cloud services may be equally irrational when it makes sense to move to more dedicated or hybrid approaches.
Pay attention to the math in Weinman's original cloudonomics post. He describes a formula for determining what percentage of an application should run in dedicated systems versus in an on-demand utility--which means there are times when at least some percentage of an application should run on dedicated systems.
There is no doubt that the caution many feel about cloud adoption is driven in part by irrational biases. However, those most in love with the public cloud concept should also be cautious that they don't avoid a good decision to own or rent dedicated systems when it makes the most sense. (Some successful Internet businesses have already discovered this. Others have already done the math to determine when it will make more sense for them to run in a dedicated environment.)
Overall, Weinman's cloudonomics work is groundbreaking. If you'll be at Structure 2010 in San Francisco later this month, be sure to catch Weinman's panel on the subject. I know I'll be there.