Whenever a new technology or methodology seems poised to shake up enterprise IT, vendors hustle to spin the phenomenon in their direction.

One such concept has been the self-managing data centre, often referred to as "autonomic computing," the term IBM favours. Here, the goal is to design and implement systems that monitor themselves, repair themselves as necessary, protect themselves from external threats and even re-route their own resources to best meet business needs. (That last factor echoes the promise of another red-hot phenomenon, on-demand - or utility - computing.)

Of course, the road to this data-centre utopia is not yet smoothly paved. But leading vendors are working hard at filling the potholes and, simultaneously, making their names synonymous with self-managing computing. We look at HP, IBM and Microsoft as they struggle to differentiate themselves and gather mind-share in the self-managing data centre.

"On-demand computing" is an IBM term that has percolated into general usage, much to the frustration of competitors. The same can be said of "autonomic computing." And savvy marketing is only part of the reason; IBM is unquestionably the industry pioneer in on-demand computing and has maintained from the start that autonomic computing is a fundamental pillar.

Analysts and users agree that Big Blue leads in the race to gain mind-share in this arena. "We've been at it longer and are focused on it more" than competitors, says Alan Ganek, IBM's vice president of autonomic computing.

Mobil Travel Guide, a division of ExxonMobil launched two years ago, chose IBM's On Demand Linux Virtual Services to host and manage e-business applications. The key reason was cost, says Paul Mercurio, Mobil Travel Guide's CIO. On-demand and autonomic computing will save the company 25 per cent to 30 per cent on IT costs in three years, he says.

The guide maintains three development environments: one for application development at its HQ, and testing and development environments on its partition of an IBM-hosted mainframe. Mobil Travel Guide applications begin life in Naples. When they're ready for testing, Mercurio's group, working with one IBM database administrator, moves them onto IBM's systems. "That's where we try to break them," Mercurio says. About once a month, the fully tested functionality is pushed into production.

Mercurio won't draw conclusions about the value of autonomic computing on its own. Any savings Mobil Travel Guide accrues come from a combination of autonomic and on-demand computing, and from buying IT resources on a pay-per-use basis, he says.

Breadth also leans in IBM's favour. IBM has "the most comprehensive approach, [building] technology all the way down to the chip and fabrication level," says Audrey Rasmussen, an analyst at Enterprise Management Associates (EMA). Indeed, IBM says more than 400 features in more than 50 hardware and software products include autonomic capabilities. IBM also is heavily invested in an open approach and actively participates in standards-setting efforts at the World Wide Web Consortium and the Organization for the Advancement of Structured Information Standards. "IBM's standards focus is unique. It's really trying to get the 'plumbing' standardised so management software can be interoperable," Rasmussen says.

HP's Adaptive Enterprise
If IBM is the first vendor that comes to mind when the topic is self-managing computing, HP is a strong second, says Dennis Drogseth, another EMA analyst. There's a key difference between the two vendors' approaches: While IBM refers to autonomic and on-demand computing as separate, if closely intertwined, initiatives, HP not only combines the two concepts in the Adaptive Enterprise, but also puts more emphasis on aligning IT resource allocation with changing business needs.

"Our programme is highly consultative. It's not a product-and-technology approach," says Ram Appalaraju, vice president of HP's adaptive enterprise programme. "All that consultation is intended to move an enterprise's people, processes and technology toward a service-oriented architecture," he adds.

This architectural approach is one of HP's "serious strengths" in the market, Drogseth says. "It has a richer investment in network management" than IBM or Microsoft, he adds, citing examples such as improved monitoring capabilities in OpenView and an increased number of automated capabilities for server optimisation. HP's "secret strength" in Adaptive Enterprise is its Internet Usage Manager, which tracks and automates the optimization of enterprise Internet use, he says.

Heading into 2005, the quick take is that IBM is the clear first in autonomic computing, and HP a worthy second.

Microsoft's DSI
Many would argue that the vendors best positioned to battle with IBM and HP are Computer Associates and BMC Software because of their enterprise management roots. But because of the massive installed base of Windows, corporations also must stay abreast of Microsoft's plans for self-managing computing.

While Microsoft once positioned its Dynamic Systems Initiative (DSI) as a true enterprise-scale autonomic computing effort, it has backed off slightly from that grand vision. What has emerged is more of an application life-cycle methodology "moving toward autonomic computing," Rasmussen says.

Michael Emanuel, a Microsoft director of product management, explains: Once DSI is fully realised, it will "look across the life cycle of an application and capture knowledge of what it takes to run a complex system. We want to make it so that humans manage by exception, while the system does all the day-to-day work."

Underlying DSI is System Definition Model (SDM), a common schema for storing configuration and performance data. In Microsoft's first step into DSI, it is building SDM into development tools, starting with the forthcoming Visual Studio 2005, Emanuel says. When a corporation uses these SDM-enabled tools, applications "will have a concept of what it means to be healthy," he says. What this means, he says, is that functions handled by a management console today will be baked right into applications themselves.

But DSI, because "it's very monolithic, for Windows only," likely won't serve as a genuine end-to-end enterprise solution, Rasmussen says. IT managers could conceivably use Microsoft Operations Manager add-ons to extend DSI to Unix systems, "but that still doesn't get you to mainframes," she says. In her estimate, DSI is more likely to catch on primarily in small to mid-sized businesses that are all-Windows shops.

Indeed, Microsoft privately concedes that its offerings can't manage other vendors' software "all that deeply," Rasmussen says. Nevertheless, DSI understands that Windows is part of something bigger - which Emanuel contends is key.