October 15, 2008

The IT forecast calls for widely scattered clouds, in the sense that cloud computing looms on the horizon.

That was the consensus among the panelists at Tuesday's Cloud Summit Executive conference at the Computer History Museum in Mountain View, Calif.

Cloud computing is coming, everyone seemed to agree, slowly to large organizations, more quickly to smaller ones, largely because it is financially compelling.

MR Rangaswami, co-founder of Sand Hill Group, presided over the event with the aim of reaching some consensus about the definition of cloud computing, providing some insight into where the clouds will be forming, and examining where new opportunities might shine.

The definition remained elusive. Cynics might prefer Oracle CEO Larry Ellison's insistence that cloud computing is marketing gibberish or free software guru Richard Stallman's assertion that cloud computing is a trap.

While any cloud computing offerings should be evaluated for gibberish and lock-in risk, such definitions are too broad to be meaningful. Every business has different needs and not every cloud computing service can meet those needs.

Columbia law student and tech blogger Luis Villa, in a recent blog post, observed that there are really four kinds of clouds: traditional applications, hosted elsewhere (Salesforce.com); services involving data that can't (yet) be managed locally (Google Maps); services that make creation of new data technically or economically feasible (Wikipedia); and services offering computing and storage, rather than data (Amazon Web Services, or AWS).

Vishal Sikka, CTO of SAP, at the conference, described for three kinds of clouds: Google App Engine and AWS-style platforms; outsourced clouds run by the likes of HP and IBM; and the clouds companies run themselves. He was there to argue that large enterprise software makers like SAP and Oracle still have a place where clouds gather and to pitch his company's vision of the SAP business suite as a single application that will eventually be able to serve any business need.

Among the descriptions of cloud computing heard at the conference, Amazon Web Services came up the most frequently. It offers the most clearly defined cloud play: a no-frills, pay-as-you-go computing infrastructure platform.

Cloud computing was described as having the following core characteristics: transparent scalability, usage-based billing, and off-premises operation.

More broadly, cloud computing can be thought of as the urge to stop worrying about how the bits flow in an organization, in order to focus on how the flow of bits can create business value.

And under that definition, Amazon Web Services falls short. As Navin Chaddha, managing director of the Mayfield Fund pointed out in a panel discussion, AWS is still hard.

In that difficulty, Chaddha sees the opportunity for new companies to help manage the cloud, to make IT disappear. If the cloud is indeed computing as a utility, as some definitions suggest, the complexities of coding and networking should be invisible to cloud customers, just as the delivery of electric power happens out of sight and out of mind of utility customers.