The Chaos of Cloud ComputingBy Michael Vizard | Posted 2008-06-02 Email Print
We need to marry the best of cloud computing with the best of what has gone before.
At this point, most people in the industry agree that cloud computing is going to play a relevant role in the future of enterprise computing. But, as with every new emerging trend, there is a tendency toward extremism that leads people to erroneously conclude that cloud computing is about to replace everything we know about how enterprise computing works today.
The impact of cloud computing—and all things Web 2.0—was the subject of a recent roundtable discussion among an independent group of chief technology officers who meet every other month to discuss technology trends. The CTOs who attended the Herndon, Va., event, which is run by Yuvinder Kochar, CTO of The Washington Post, viewed the hype surrounding Web 2.0 with a fair amount of skepticism. They all agreed that cloud computing would play a role in enterprise computing, but the consensus was that it would play a supporting role rather than a starring one.
The key limiting factor in the mass adoption of cloud computing was summed up best by Dan Geller, CTO of Technology Services for Legg Mason, a financial services company. Traditional business-to-business companies have developed hundreds of thousands of lines of application code—code that was never designed to run as a service. Most of them are still working out how to build applications using a service-oriented architecture, which might allow them to outsource some applications to run as a service in a cloud. But the vast majority of their mission-critical applications would not be able to run as a service.
The consensus of the panel—which also included Jeff Barr, senior Web services evangelist of Amazon Web Services; Johnny Barnes, vice president and CTO of IBM’s Public Sector Group; and Vijay Manwani, executive vice president and CTO of Bladelogic, which is being acquired by BMC—was that the difference between cloud computing as discussed today and traditional outsourcing comes down to scale and the fact that the compute resources will ultimately be made available dynamically to customers. For example, Amazon started out offering storage services on demand to customers and is now partnering with Sun to offer compute cycle on servers on demand.
Coincidentally, the folks at Accenture have also been studying the strategic role cloud computing will play in the enterprise and have concluded that we’re faced with a paradox. On one hand, says Accenture Chief Scientist Kishore Swaminathan, companies should take advantage of cloud computing to deliver commodity-level services, whether they involve additional storage or server capacity or on-demand applications. On the other hand, cloud computing provides a unique opportunity to fast-track strategic business initiatives.
First, cloud computing provides an IT platform to support those initiatives without requiring a major investment. This means companies can afford to invest in a new business opportunity with a limited IT outlay. Second, it provides a platform that can scale quickly should the business opportunity prove wildly successful. Swaminathan says this creates a paradox because cloud computing as an IT platform is best suited for both your least valuable IT activities and your most strategic business initiatives.
None of this means that the enterprise’s mountains of on-premise applications are going away any time soon. What IT has to get its arms around is how to blend cloud computing services from a host of providers with internally managed applications to create a more flexible IT environment that can better support the dynamic needs of the business.
Despite the extreme positions some people are taking about cloud computing—outlined in Nicholas Carr’s latest book, The Big Switch—common sense dictates there is not going to be a wholesale migration to a new computing model that ignores everything that has gone before.
Extreme statements help sell books, but any strategy based on book-of-the-month-club management is doomed to fail. What we need is a reasoned approach to enterprise computing that takes advantage of the best of what new computing models can offer and marries it with the best of what has gone before.
The fact is that no new computing model ever completely replaces an older one. Instead, what we see time and time again is how new computing models are slipstreamed into the enterprise computing architecture to give IT organizations the richest set of options available to support the business.