News

Consultant: Grid's the 'next big thing'

• By John K. Waters
• July 1, 2002

Grid computing, the collaborative, network-based model that enables the sharing of data and computing cycles among many processors, represents a new computing paradigm with the potential to dramatically alter the IT industry's competitive landscape.

That's the conclusion of a new report entitled ''The Global Grid Computing Report 2002 -- Technology and Market Opportunity Assessment'' from consulting firm Grid Technology Partners. The report was unveiled during the recent Grid Computing Planet Conference & Expo in San Jose, Calif. ''Companies that pay attention to this evolving technology will stand to reap huge gains in the coming years,'' the report declares.

Demand for computing power in industries such as the life sciences and financial services is virtually unlimited, said Ahmar Abbas, managing director of San Jose-based Grid Technology Partners (http://www.gridpartners.com ). Because the grid computing model leverages existing resources, it delays the need to purchase new infrastructure, and presents an extremely compelling picture to budget-strapped IT departments.

''Grid is cheaper than a server farm,'' Abbas told conference attendees. ''You can leverage whatever you have and avoid the wholesale replacement of infrastructure.'' Abbas, who worked at UUNET and ONI Systems as an electrical engineer before launching Grid Technology Partners, said that he has talked with a ''big Wall Street firm'' that was willing pay $1 million for a server farm just to cut the time it takes to run risk calculations in half. With grid computing, the time could be cut even further, he said, and at much lower cost.

According to the company's debut research report, ''Grid Computing takes collective advantage of the vast improvements in microprocessor speeds, optical communications, raw storage capacity, World Wide Web and the Internet that have occurred over the last five years. A set of standards and protocols are being developed that completely disaggregate current compute platforms and distribute them across a network as resources that can be called into action by any eligible user (person or machine) at any time. A company with 600 grid-enabled desktop PCs can utilize all of them together as one compute platform -- suddenly providing it with enough computing capacity to go head-to-head with the world's 49th largest supercomputer.''

But the advent of the grid isn't good news for everyone, Abbas said. For example, even though Sun Microsystems is often listed among the leading firms in this space, the company is ''at risk'' because its ''software and hardware platforms are tied too closely together,'' he said. IBM, on the other hand, is ''on the mark'' because of its more neutral approach to software. IBM's work on the Globus Project, a multi-institutional research and development venture that is creating fundamental technologies for computational grids, in particular sets the company apart in this space, Abbas said.

A primary product of the Globus Project is the open source Globus Toolkit, which is being used in numerous large grid deployment and application projects in the United States, Europe and around the world. The Globus Project is based at Argonne National Laboratory and the University of Southern California's Information Sciences Institute.

Another sign of increasing interest in the grid-computing model, Abbas pointed out, is the top-seller status of a grid textbook edited by Globus team leaders Ian Foster and Carl Kesselman. According to Abbas, it is one of Amazon.com's biggest sellers to readers in Menlo Park, Calif., the epicenter of venture capital.