Home > Applications

Computerworld - CIOs are constantly challenged to stretch IT resources on limited budgets, leaving few assets for innovation. But what if IT systems could be allocated based on actual use and need?

That's the idea behind Hewlett-Packard Co.'s Tycoon, a virtualized, market-based system for delegating computer resources. Tycoon pulls a company's IT resources into an abstract utility layer. Users can bid or barter for computer resources within their own companies or with other companies in a broader market.

Hewlett-Packard Co. http://tycoon.hpl.hp.com PRODUCT:  Tycoon DEVELOPERS: Bernardo Huberman, Kevin Lai and economists and computer scientists at HP’s Information Dynamics Lab

Within a company, employees are given a certain number of tokens per year to spend on computing resources. The more urgent a user deems the need, the more tokens he can bid for use of the equipment. In a broader market, resources could be offered online to the highest cash bidder.

While the concept of shared resources isn't new, HP's convergence of the principles of two disciplines, economics and technology, to develop Tycoon is.

"Part of the reason why it has taken so long for it to get off the ground is that economists lack the expertise to [build a system] . . . and computer scientists lack the understanding of incentives and organizational structure to make a system that can do these kinds of things," explains Kevin Lai, a scientist at HP's Information Dynamics Lab in Palo Alto, Calif., and a key developer on the project.

Bernardo Huberman is credited for bringing the eclectic group together back in 2004. "You need [economists] interested in solving concrete and real problems," explains Huberman, senior HP fellow and director of the Information Dynamics Lab. "Also, you need people from computer science with enough vision that they want to do something different."

But it was Lai who mediated between economists and computer scientists and even intrigued visiting scientists who heard the buzz about Tycoon and wanted to offer their input.

For example, Lars Rasmusson, now at the Swedish Institute of Computer Science, and HP research scientist Li Zhang heard about the project and contributed algorithms for Tycoon.

The relationship between the two disciplines wasn't always harmonious. Lai found that many economic theories didn't apply in practice. "Previous work from economists didn't take into account the realities of computer systems," he says. "Like the fact that they fail, there is latency, and that users put a high value to having low latency to do these jobs quickly. We had to take into account