Cloud Computing: Been There, Done That?

Cloud computing may be the next big thing, but it didn't happen overnight. Its genealogy goes all the way back to time-sharing, a practice dating to the 1960s in which a mainframe computer was accessed simultaneously by multiple terminal users in different locations via telephone lines.

Cloud computing also resembles the ASP (application service provider) phenomenon, which cropped up a decade ago as the Internet and Web were taking the world by storm. ASPs were quickly joined by a host of other Internet-based services, including managed service providers, storage service providers and other "XSPs" that went through a rapid boom and bust cycle.

Surviving ASPs such as Salesforce.com began calling themselves SAAS (software as a service) vendors, which some analysts include in the cloud computing universe.

In addition to its flagship CRM service, Salesforce.com recently launched a cloud offering, Force.com, that the company calls PAAS (platform as a service). Force.com provides a platform for application development on demand.

Another ancestor to today's computing clouds is grid computing, a model in which large numbers of servers are linked together in a kind of fabric in one or more locations. Some clouds use grids to deliver their services, while others do not. Sun Microsystems, for example, has been offering up CPU cycles for computing-intensive tasks on its servers through its Sun Grid Compute Utility and Network.com Web site for several years at the rate of $1 per CPU per hour.

Meanwhile, IBM has been touting "on-demand" computing and Hewlett-Packard has been offering "utility" computing services since the early 2000s. Both models are geared to augmenting data center needs of large companies with off-premises capacity.

"Cloud computing is the next big label for something that's been going on for a decade," said IDC analyst Frank Gens. "There seems to be critical mass building right now for this phenomenon. This has been an ongoing development of IT offerings, business services and consumer services."

The key differences between these various precursors and today's clouds, observers agree, is a self-service Web customer interface, combined with the ability to rapidly add or subtract computing capacity.

While previous services, such as those from Sun and IBM, usually required some handholding for a customer to get started, most cloud services can be fired up by a customer wielding a credit card in a matter of minutes from a browser interface.

Further, the ability of cloud services to expand or contract quickly and with little or no notice is typically enabled by virtualization technologies, which can create any number of virtual machines as needed.

If cloud computing does take off, it may prove a theory propounded by controversial industry observer Nicholas Carr, whose most recent book, "The Big Switch," covers the emergence of Internet-based computing services, including clouds.

Carr's previous volume, the widely disputed "Does IT Matter?" laid out his theory that corporate computing is rapidly becoming a generic commodity, much like electricity.

At one time, businesses generated their own electrical power with their own equipment. Over time, electrical power became a generic commodity accessed, as today, over the power grid. So, Carr contends, cloud computing is a natural evolution from premises-based IT operations, and, over time, nearly all our computing services will come from a cloud.

"I think it carries out my thesis," said Carr in an interview. "I do think it's the next big thing—it's where the innovation is today in computing. More and more, what used to be done in-house will shift to the cloud. That doesn't mean it will happen in one, two or five years. For big companies, it will take a long time. It does seem to be playing out the way I anticipated. In the past four years or so, I've been surprised at how quickly things have moved forward."

IDC's Gens agreed regarding the long-term significance of the cloud phenomenon. "Cloud-enabled computing is the model for the next 20 years of the IT marketplace," he said, "but it's not here yet. Timing will be everything."