Grid-lock?

Having already given us the World Wide Web, CERN (the European Organization for Nuclear Research), is brewing a prime example of what could turn out to be the next revolutionary advance in computing: the grid. Grid computing, for the uninitiated, is a form of distributed computing that dynamically coordinates and shares processing, application, data, storage or network resources from multiple linked computers.

Last week, UK particle physicists demonstrated the world’s largest working grid at the e-Science All-Hands meeting in Nottingham. With over 6,000 computers at 78 sites internationally, the Large Hadron Collider Computing Grid (LCG) is the first permanent, worldwide grid for science experimentation. The particle physics experiments at the LCH, which is under construction at CERN in Geneva, are expected to produce some 15 petabytes of data per year. By 2007, the LCG 'virtual supercomputer' will be equivalent to 100,000 of today's fastest computers working together, the scientists said. This map shows, in real time, how computing jobs move around the LCG as it stands today.

Meanwhile, grid computing for enterprise applications is making progress. Commercial grid applications today include reservoir modelling for petroleum exploration, actuarial analysis in insurance industry, circuit simulation at electronics firms and structural analysis in aerospace and automotive companies. Several companies -- including Avaki, Data Synapse, Platform Computing and United Devices -- specialise in grid software and have thriving businesses serving Fortune 1000 companies.

In addition, more mainstream applications are becoming grid-friendly. For example, Platform Computing has an adapter for Microsoft Excel that grid-enables compute-intensive worksheets. Oracle 10g enables databases and applications to be run on grids, improving performance and increasing hardware utilisation.

So called 'extragrids' will enable enterprises to share computing resources with trusted partners and suppliers in a kind of federated grid. The final stage, which will take a decade to evolve, is the utility grid, in which computing resources are delivered on-demand in a pay-for-use model.

Interested parties have already formed several grid-related groups: the Global Grid Forum (GGF), whose major backers include IBM and Microsoft, is broadly based, while the Enterprise Grid Alliance (EGA), which was launched this April and includes Oracle and HP among its luminaries, is more focussed on businesses and datacentres. Then there's the Globus Alliance, which is more concerned with science and engineering.

This is all good news, but only so long as the proliferation of industry organisations doesn't hinder rather than help the development of protocols and standards in this potentially world-changing arena.