Analysts Weigh Strategies of On-Demand 'Hydra'

By Clint Boulton

(Back to article)

Representatives of research firm Summit Strategies Tuesday outlined the virtualization aspect of on-demand computing, and weighed the pros and cons of HP's, IBM's and Sun Microsystems' respective strategies for that ideology.

The analysts took time to discuss how virtualized computing -- the practice of automating consolidation and pooling of IT resources in a way that allows them to be managed as a "pool" of server, storage and networking capacity that can be allocated on the fly -- may be used by IT customers to save money. Virtualized computing fits under the broader umbrella of utility computing, according to the analysts.

The analysts also examined how the major competitors stack up against one another right now. Summit Strategies President and Chief Research Officer Tom Kucharvy and Senior Analyst John Madden took time to explain that there are a confusing number of terminologies for the notion of utility or on-demand computing, which is the idea of computing resources that are piped as needed to accommodate complex, evolving businesses.

Madden pointed out that every major vendor has different names for it: IBM loosely uses the term on-demand computing but describes its framework as the Utility Management Initiative; HP calls its strategy "adaptive infrastructure," but their broader canvas is the Utility Data Center; Sun calls their strategy N1; and newcomer Microsoft uses the phrase Dynamic Systems Initiative to describe the harnessing of computing resources on the fly.

Currently, IBM, HP and Sun are all offering some of these characteristics from their respective plans. IBM features many servers with virtualization (including Linux-oriented zSeries mainframes); HP features a virtualized server; and Sun recently rolled out virtualized blade servers under the auspices of N1.

Kucharvy detailed strengths and weaknesses he could discern among HP's, IBM's and Sun's utility computing strategies, calling the sector largely "hydra-headed" because, though other firms are embarking on similar ventures, those three are leading the race.

"Sun's N1 has a nice technical vision. Overall, it is most compelling for corporate CTOs," he said. "However, it has limited services, software, a late start. There is also the question of whether they can make the huge financial investment required."

Kucharvy said HP has the technology skills in place, as well as the business skills and ability to measure ROI, but that it's marketing prowess is weak in this vein and it tend to overrely on partners. Still, UDC is attractive to CIOs and HP's hidden trump card may be its OpenView software platform, which could give them strong market share in utility computing.

IBM, Kucharvy said, has the distinction of making on-demand ubiquitous throughout its entire corporate strategy. He said IBM has the integration strategy, reputations, marketing, software, services, and ISV program to get it done, but that its vulnerability is trying to be everything to everybody in the bet-the-company strategy might be too "mushy." This on-demand foray, he said, is attractive to CEOs.

But there was also discussion of Microsoft's new Dynamic Systems Initiative, which Kucharvy said is not only late to the game, but extremely vague.

"Microsoft's DSI has got a lot of components, they are trying to integrate new management tools, and really support it through Visual Studio .NET," Kucharvy said. "But Microsoft's is late to market and seem to be falling into the same problem they did with .NET two years ago of not really outlining a clear strategy. Microsoft will need to take explain their vision to partners, and get something to market."

Speaking of ambiguity and confusion, Kucharvy said: "There is a desperate need for a standards lexicon. Unless a major vendor tries to impose one, it will be hard to see how customers adopt [utility computing]." He suggested IBM might have the wherewithal to tackle this issue.

Madden echoed that, noting that proprietary lock-in from the lack of standards to go along with the terminology is another chief detractor from keeping this ideology of on-demand from reaching full fruition.

Madden also explained that the idea of virtualized computing is not new, having long been used to describe a form of data management in storage, or even IBM's logical partitioning capabilities in its servers: the practice is, however, expanding to data centers.

Madden said virtualized computing currently addresses pain points that haunt IT companies in a financially-constrained economy by cutting into IT capital costs and operations costs; improving poor reliability and availability; and limiting operations disruption. One obvious example of the cost-cutting is a virtualized server that does the job of multiple servers.

Madden said the goal of these firms is to automate the computing process, and then make them as simple as possible. To do this, he said, virtualized offerings will have to work across various systems, and there will have to be heterogeneous products from vendors. At present, Madden said there exists only virtualization internally, for their own platforms. This will have to change, he said, to ensure openness, thereby staving off vendor lock-in.