Cloud computing appears to be in that nebulous stage that all new technologies go through as they move from the lab into the public domain: Everybody's talking about it but no one is really sure what it is.
The broad outlines are in place. Cloud computing basically provides a means to deliver all data center resources -- processing power, storage, applications and data -- over a network, providing enterprises with an entirely new level of flexibility in delivering services to users, or even hiring them out to customers as Amazon and Google are doing.
That's all well and good as far as it goes, but when you get into the details, the simple explanation starts to get more complicated.
The first, and probably most obvious, question is how is this any different from the utility computing movement of just a few years ago. Andreas Viklund at IT Spot posed that question to a number of insiders and concluded the main difference is that virtualization both lowers the cost and greatly enhances management capabilities to truly deliver on utility computing's initial promise: to make data center resources available like electricity, delivered over a network and paid for according to usage.
One of the companies stressing the management aspects of the cloud is Elastra Corp., which recently introduced the Elastra Cloud Server with an eye toward simplifying the launch of clustered database applications for on-demand environments like Amazon's Elastic Compute Cloud (EC2). The system provides a design function to oversee application set-up and implementation, coupled with a run-time function that connects to the cloud, allocates appropriate virtual resources and installs the database and related software, instrumentation, metering and management tools.
But if your enterprise is looking to establish a cloud either for internal use or to lease out to clients, how will you know if you really have one? Analyst James Governor offers a not entirely tongue-in-cheek checklist here, listing 15 ways to tell it's not cloud computing, done up in the old "You know you're a redneck..." fashion. My personal favorite: "If you know where the machines are... it's not a cloud."
In the view of Irving Wladawsky-Berger, chairman emeritus of IBM's Academy of Technology, many data centers are on an evolutionary path toward the cloud already, with virtualization as only the latest mutation in the process. The result will be a system-wide distributed computing network, but the catch is that the same old problems you face today will haunt you in the cloud: how to manage resources, how to govern data exchange, how to simplify operations ...
If you look back at nearly all the major IT innovations of the past 30 years, few were able to define themselves fully until well past their introductory phase. The same is likely to happen with the cloud. And even though vendors and experts will render their take on what it is and what it can do, the real definition is likely to come from the user community.