Big fix for a big problem
- 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
Grid is a long-term fix for the big problems of IT, says Information Age editor Kenny MacIver.
It is an unprecedented situation when the warring factions of the computer industry embrace a single vision of the future. But over the past year, if you ask the question of where the industry is going of IBM, Hewlett-Packard, Sun, Dell, Oracle, Veritas, Computer Associates and many others, one answer comes back: grid computing. They may couch the answer in terms of the steps to that model of distributed computing - consolidation, on-demand, pay-as-you-go, utility data centres, virtualisation - but the aim is the same: of delivering computing as a flexible, efficient, dependable resource that can draw on a network of computing power and so scale to meet fluctuating demand.
The early implementations of grid were almost entirely dismissed by many in IT as only suitable for scientific and technical computing. But now the model is making its way into the corporate arena. To see names such as Abbey National, Novartis, Morgan Stanley, Nomura Insurance, Philips Semiconductor and Hewitt Associates starting to shout about how grid and utility projects are helping them make much better use of their existing resources shows how far grid technology has come in 2003.
At this point, almost all of these applications are similar in structure - they tackle compute-intensive problems by splitting processing into discrete parts and executing these on low-cost, under-utilised machines distributed across a network.
Pharmaceutical company Novartis, for example, says it has avoided the expense of buying an additional supercomputer by setting up a grid that draws on the unused processing cycles of 2,500 PCs to run a drug discovery application.
Our cover line asks the question: 'True grid: Today, tomorrow or a decade away?'. The answer is all three. The initial implementations of grid are already there, if narrow in scope. Over the next three to five years, the model will start to be used for mainstream applications. But it will take the best part of a decade before a utility-like capacity is available.
Organisations are rightly sceptical. But for once, the computer industry is aiming at a higher target than the next killer app or the next hot box. It is trying to leave behind the inefficiency, the inflexibility of a discredited model and supplant it with one that is much more responsive and adaptable to the needs of the dynamic needs of the organisation.
