© 2009 Bernard Golden
EDUCAUSE Review, vol. 44, no. 4 (July/August 2009): 64–65
The system was built and installed. After being rolled out, its user population was rapidly increasing due to its social media characteristics and undoubted user value. But it had some weaknesses as well, rooted in its physical infrastructure. The system ran on one machine in a hosting center. The system sponsor — along with the entire user population — was exposed to downtime should anything break. Although the system owner paid for maintenance, there was no system redundancy, meaning that it could take hours or even days to recover from a hardware failure.
Does this dilemma sound familiar? Eventually the application was migrated to a cloud provider. With a relatively small amount of engineering, the system was soon running in Amazon's cloud infrastructure — with easy recovery, protection against hardware failure, and as an added benefit, redundant software components providing high availability. The best part? The total monthly cost came to less than one-half of the previous hosting fee.
Stories like these have raised the interest in cloud computing to what one analyst recently characterized as "DEFCON 2 on the hype scale." With so much interest, one might be tempted to dismiss cloud computing as a fad. But is it?
Most of those in the computer industry don't think so. IBM, Microsoft, VMware, Sun Microsystems, and a host of others are poised to invest literally tens of billions of dollars in cloud computing. If this is a fad, it is unprecedented in the amount of money and the number of leading vendors involved.
So, what exactly is cloud computing? Although it seems that every vendor (and indeed, every person) has a definition of cloud computing, I like the one proffered by the Electrical Engineering and Computer Science (EECS) Department at the University of California, Berkeley, in "Above the Clouds":
- The illusion of infinite computing resources available on demand, thereby eliminating the need for Cloud Computing users to plan far ahead for provisioning
- The elimination of an up-front commitment by Cloud users, thereby allowing companies to start small and increase hardware resources only when there is an increase in their needs
- The ability to pay for use of computing resources on a short-term basis as needed (e.g., processors by the hour and storage by the day) and release them as needed, thereby rewarding conservation by letting machines and storage go when they are no longer useful1
This limns the definition of an agile infrastructure: resources ready when you are, no onerous barriers to getting started, and cost associated directly with use, with the ability to dial total resource consumption up or down as needed.
The poster child for enterprise cloud computing use is the New York Times. Having digitized over a century's worth of newspaper pages, it needed to convert them to browser-friendly PDF format. After examining whether the job could be done internally — it could, but it wouldn't get started for a while and would require the purchase of more hardware — one engineer decided to give Amazon Web Services (http://aws.amazon.com/) a try. He set up twenty compute instances and ran the entire 4 Tb job over a weekend, whereupon he shut down the instances. Total cost? $240, which he paid with his personal credit card.
Examples like these bring home the potential of cloud computing. On the other hand, many IT pros have reservations about cloud computing — not about the premise or even the promise, but about the scope of applications for which cloud computing is appropriate. They recognize that every new platform brings with it challenges that must be addressed — and solved — to achieve maximum benefit.
What are the issues inherent in cloud computing? Most institutions face three types of challenges in leveraging cloud computing: practical issues, political issues, and policy issues.
- Practical Issues: These are the "How?" questions. How do we know which cloud platform is right for the institution (or, indeed, should the institution be focusing on building its own, internal, cloud)? How can staff get up to speed on the new development frameworks and practices of the chosen cloud provider? How can existing system management products and processes be integrated with the cloud?
- Political Issues: These are the "Who?" questions. Who gets to decide what to move to the cloud? Who gets to set the decision criteria? Who will be responsible for the cloud systems? Every new technology development benefits some people, whose jobs are made easier. Those people embrace the development. But every new technology development threatens other people, whose functions are less important in the new environment. Those people resist the development. For example, the manager in charge of data center operations might not look too favorably on using external cloud providers, but a business IT service user is likely to eagerly pursue the rapid resource availability of cloud environments.
- Policy Issues: These are the "What?" questions. What data can be stored in the cloud? What laws or regulations govern what can be done with data processed in the cloud? What can an institution do to assess which systems can legally be moved to cloud environments?
Answers to these questions can be found by understanding and addressing each type of challenge:
- Practical issues, which commonly arise when people are getting up to speed with new technologies, accompany every platform shift. As an institution gains familiarity with initial cloud systems, as staff go through educational sessions, and as new employees join the institution, these issues fade. In one sense, they are human capital issues that are naturally addressed through time and experience. A good approach to working through this type of challenge is to set up a pilot program to explore the practical issues. Learning can occur via analysis and debriefing after the pilot program ends.
- Political issues can be more troubling. In the past, many useful technology initiatives have been significantly hampered and delayed due to infighting. For cloud computing, these issues will likely be transitory. The financial benefits of cloud computing are so evident (à la the New York Times example above) that opposing it is self-defeating. When the evidence is clear-cut, no one wants to be on the wrong side of the argument. The right way to address this type of challenge is to attach a financial analysis to the pilot program and to subsequent cloud projects.
- Policy issues are perhaps the most troubling, due to the fact that resolution is outside the control of any one institution. I expect that there will be confusion and questions for all institutions as they consider their most sensitive systems with respect to the cloud. For those systems, arranging for consultation with and guidance from legal counsel is appropriate. However, the mistake made by some institutions — dismissing cloud computing as a whole because some systems are governed by laws or regulations — should be avoided.
Compared with previous platform shifts, such as client/server or the Internet, cloud computing is attracting interest and enabling uptake much more quickly. This is the result, I think, not only of a dissatisfaction with the agility of information technologies but also of the growing recognition that the cost and the complexity of data centers are getting out of hand for most institutions. Unlike fads, cloud computing is unlikely to fade away. A good response is to heed the Boy Scouts' advice: "Be Prepared."
- Michael Armbrust, Armando Fox, Rean Griffith, Anthony D. Joseph, Randy H. Katz, Andrew Konwinski, Gunho Lee, David A. Patterson, Ariel Rabkin, Ion Stoica, and Matei Zaharia, "Above the Clouds: A Berkeley View of Cloud Computing," EECS Department, University of California, Berkeley, Technical Report No. UCB/EECS-2009-28, February 10, 2009, p. 1, <http://www.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-28.pdf>.
The author talked in more detail about the policy implications of cloud computing in an EDUCAUSE podcast interview: