There's a lot of gnashing and grinding of teeth over cloud computing these days. Depending on whom you talk to, it's either the second coming of enterprise computing or a potential security threat to all we hold near and dear.
Recent surveys show that IT people are all over the spectrum when it comes to cloud computing. Yet, if we take a giant step back from all the hyperbole, some pretty obvious things start to emerge in terms of determining the immediate value of cloud computing.
The first is backup and recovery. Clearly, it is more efficient to make use of a cloud computing service to provide backup and recovery. Investing in a ton of duplicate infrastructure makes no sense.
The second area is also pretty obvious. Software-as-a-Service (SaaS) is a form of cloud computing. Whether it's a payroll or customer relationship management (CRM) system, there are times when delivering those applications as a service makes sense. A lot of times, the internal IT organization does not have the expertise required to run a particular application or that application may not be strategic enough to justify committing limited IT resources to managing it. There's no doubt there are potential security issues when it comes to cloud computing, but like all things in life the risks need to be weighed against the potential benefits.
Arguably, the next big thing in cloud computing will be more specialized application services. A lot of IT organization can't afford to invest in supercomputer-class infrastructure. Yet, the business could benefit from access to some pretty compute-intensive analytic applications. An interesting example of how one of these services might work is WolframAlpha, a query service based on a new model for doing sophisticated searches and calculations. WolframAlpha counts a company called Xignite, a provider of financial information via standard Web services protocols, as one of its partners. User of the WolframAlpha system are able to leverage the Xignite data alongside other data to query the WolframAlpha system about any calculation they can think of.
Obviously, building an equivalent to the WolframAlpha system would be cost prohibitive for most organizations and you can image how any number of compute-intensive application services might spring up on top of any number of public cloud platforms.
None of this means that on-premise applications and infrastructure are going away. On a practical level, there are far too many existing applications that can't be cost-effectively rewritten to run on a public cloud. On a strategic level, there are hundreds of applications that are too fundamental to the business to run on a cloud. And finally, there are a number of legal and regulatory issues that may not make cloud computing practical in some cases.
Cloud computing is not an all-or-nothing proposition. What we are slowly migrating toward is a blended computing model that will combine the best elements of public cloud services with on-premise applications that will run on internal IT systems that use the same architectures as public cloud services. And once that happens, we'll enter a new era of IT flexibility that should for the first time really allow IT organizations to dynamically respond to the rapidly changing needs of the business, versus always trying to get the business to conform to the way IT works.