Enterprises Still Not Sold on Grid

Despite plenty of vendor hype, it appears that most enterprises are still to be won over by the concept of grid computing, citing cost and service level uncertainty as key concerns. (See IBM Debuts Grid, IBM Brings Autonomics to Grids, Sun Grid Goes Live, and Sun Intros Grid Storage.)

Last week, for example, IBM cranked up its own grid strategy, doubling the capacity of the SURAgrid research initiative to 10 trillion calculations per second by deploying its System p servers at college campuses across the U.S. (See IBM Supercharges Grid.) Some 27 universities are now involved in the project, which harnesses their compute power for storm modeling and genome sequencing.

"This is a testbed for much broader, wider adoption of grid across commercial enterprises," says Ken King, vice president of grid computing at IBM. "This is bleeding edge from the perspective of connecting so many different organizations together in an open standards model."

Specifically, SURAgrid is using open source middleware from industry group Globus to knit the disparate sites together and share data and applications across the grid. (See Vendors Form Globus Consortium, Grid Startup Hits the Source, and Grid Goes Open Source.)

King admits that grids are still typically the preserve of the research sector, although he predicts that more and more businesses will look to the technology, particularly in data-intensive industries such as financial services and manufacturing. "Having example references of how it has been adopted in the university space makes it easier to drive adoption in the commercial sector," he says.