The search for the holy grid
- Reduce text size Decrease text size
- Increase text size Increase text size
- Print article Print
- Jump to comments Comment
- Share this article Share
- Email article to a friend Email
Is utility computing stuck on the grid?
Is utility computing stuck on the grid? That was the question that many people were wondering back in the summer after a study from IDC suggested that customers were uninterested and that it would be at least five years before the new IT architecture caught on.
But perhaps the big early leaders, Hewlett-Packard (HP) and IBM, just lacked a spokesperson who could give the idea some poke. Enter Larry Ellison and Scott McNealy, Silicon Valley's most quotable notables. The two men made utility and grid computing the centrepieces of their companies' respective autumn conferences, with Ellison, in particular, in aggressive and visionary form.
"I believe this is the dawn of the information age, not the end of it," said Ellison, introducing Oracle 10g, the company's database product designed to work across hundreds of servers. He even claimed a customer - Electronic Arts, the games software distributor.
Much of the rest of the IT industry seems to agree with that sentiment. HP and IBM are both continuing their huge development and marketing programmes, while Sun Microsystems has just completed its third acquisition in the area in the last 12 months, buying application-level provisioning specialist CenterRun for $60 million. (Earlier it bought storage provisioning specialist Pirus and server provisioning specialist Terraspring.) And storage management software company Veritas has bought two companies as it heads towards a virtualised future.
Dozens of small companies are also involved in utility computing, grid, virtualisation and provisioning - among them United Devices, Entropia, Avaki, DataSynapse and, from the UK, Oxford-based Sychron. Most have generous venture capital backing.
But there is a limit to the consensus. Microsoft is not invited to the party, because, it is said, its software is too client-centric and doesn't cluster sufficiently well. And IBM and HP rely too much on big computers - according to both McNealy and Ellison, at any rate. Oracle and Sun, for their part, are both focusing, at least in their marketing, on the benefits of arrays of smaller machines. "The price/performance difference [centralised Unix or mainframe servers versus arrays of smaller devices] is 30 to one. You have to be really willing to pay more to run slower," says Ellison.
Meanwhile some visionaries are starting to ask a more fundamental question: if electricity grids give us massive, state-wide or city-wide black outs, why won't computing grids do the same?
