IBM’s Palmisano Touts a Shift in Strategy

Big Blue’s CEO makes the case for an “on-demand” computing infrastructure

• By Stephen Swoyer
• 11/07/2002

When all was said and done, the technology vision that IBM chief executive officer Sam Palmisano outlined last week didn’t herald a raft of new products from the world’s largest technology company. Nor did Palmisano’s Computing-on-Demand announcement immediately disclose the availability of significant new IBM service offerings. In fact, Palmisano didn’t have very much in the way of new products or services to trumpet.

Instead, he was selling a vision.

“What we’re seeing is a fundamental shift that as a long-term shift is irreversible,” he suggested in an address to IBM customers in which he introduced a strategy&$151;best described as “on-demand” computing&$151;that he says will transform the way in which computers and computing resources are bought and sold. “The on-demand business fundamentally requires a different approach to how we borrow, design and build systems.”

As an alternative to the conventional build-and-buy approach in which customers purchase new systems or acquire additional storage capacity to meet increased demand, Palmisano described a utility-like approach to the allocation of computing resources in which&$151;at some point&$151;IBM or its partners could offer customers computing, storage and network resources as hosted services.

The vision Palmisano outlined is really an agglomeration of technologies that IBM has offered, in some cases, for years. First, on-demand computing borrows heavily from so-called “grid computing” (in which application software facilitates the distribution of workloads to under-utilized compute resources) as well as from IBM’s project eLiza “autonomic” computing technologies. Not surprisingly, Palmisano’s vision of an on-demand computing infrastructure also encompasses the full range of IBM’s eServer product line, and explicitly identifies the Linux operating system as an “open standard” for on-demand computing. Finally, and least surprising, was Palmisano’s acknowledgment that IBM’s new strategy will heavily leverage the expertise of its services units.

A cynic might charge that Palmisano is selling vaporware. But this is IBM, a company that markets a bevy of hardware and software products that are ubiquitous throughout a range of embedded appliances and back-end mainframe environments. Besides, in 1996, when CEO Lou Gerstner described a vision for a convergence of Internet technologies and business operations&$151;he called it “e-business”&$151;Big Blue didn’t have much to offer in terms of new products or services either. Today, notes Rob Enderle, a fellow with consultancy Giga Information Group, IBM is virtually synonymous with e-business.

“IBM presented itself, even before they had a solution, as the leader in e-business, and they basically created that wave themselves,” he points out. “IBM isn’t the first to do this, but more so than any of their competitors, they’ve got the credibility to actually go out and do it.”

To be sure, IBM isn’t alone in advocating a view of an increasingly flexible and intelligent computing infrastructure. Sun Microsystems Inc., for example, recently described a new initiative, dubbed N1, that focuses on tying together and commoditizing computing, storage and network resources to maximize overall resource utilization.

“N1 looks at … how do you make it more efficient to administer the environment, which addresses the people cost by automating and aggregating and centralizing the control,” explains Sun vice-president of marketing Andy Ingram. “Secondly, it addresses how to achieve higher resource utilization rates in the environment.”

In this respect, acknowledges Richard Partridge, vice-president of enterprise servers with consultancy D.H. Brown Associates, the infrastructure described by Sun’s N1 is similar in many ways to IBM’s on-demand vision. As if that’s not enough, Partridge points out, Hewlett-Packard Co. is also pursuing a similar approach with its Utility Data Center and Adaptive Computing initiatives. Partridge argues that such comparisons miss the point, which is that on-demand computing doesn’t purport to be a “patentable, unique, no-one’s-ever-thought-of-this-before concept. I don’t think it’s supposed to be that. I think it’s a recognition that there are waves or phases of the way computing is used, and that the flexibility to redeploy as business needs change … is a great advantage”

That said, IBM’s vision of on-demand computing differs fundamentally from N1 and other competitive solutions, Enderle says, because of the array of products, technologies and services that IBM can today muster in support of it. “IBM has done a particularly good job translating [what it means by on-demand computing] in a way that customers can understand. For example, [an on-demand infrastructure] would be very services intensive, and IBM has arguably the largest services organization on the planet. Customers see that IBM isn’t … promising something that they can’t deliver.”

As for the technology components that will comprise the nuts-and-bolts skeleton of its on-demand business strategy, Palmisano touted the success of a grid computing project that IBM developed in conjunction with the University of Pennsylvania. Grid computing, he stressed, “is here.” Palmisano also sketched a vision in which IT organizations will one day be able to exploit the resources of computers distributed across the Internet if they require additional processing power or storage capacity. IBM’s project Eliza, an initiative by which Big Blue has incorporated autonomic (i.e., intelligent self-healing) features into its hardware and software products, will support this and other capabilities.

Palmisano's emphasis on grid computing doesn’t mean that Big Blue’s other platforms will be at a disadvantage, says D.H. Brown’s Partridge. Rather, he suggests, on-demand computing as championed by IBM describes a flexible set of technology and service offerings that can be brought to bear to solve particular business problems. “This isn’t to say that all computing is the same. There are some different kinds of computing, and there are times that you will need the capabilities of mainframes, and there are times you will need grid computing. If you can shift workloads around and change the way that you distribute [workloads], then you have a way to reconfigure your computing infrastructure and your computing demands.”