Researchers set up European computer grid

CERN, the famed Swiss high energy particle physics lab, has a problem

CERN, the famed Swiss high energy particle physics lab, has a problem. It is about to start generating more data than any computer system in the world can analyse. The prospect has led CERN to drive a major European project to create a vast research network of computers across Europe, in which the Republic will participate through Trinity College Dublin (TCD).

When completed, the €10 million (£7.87 million) endeavour, called DataGRID, will become a principal European computing resource for researchers in many disciplines.

Grid computing - high-powered clusters of smaller computers - is currently the Liz Hurley of the computing world, considered very sexy indeed. Grids are a favoured topic when researchers in the big data-generating fields such as physics, biology, astronomy, and environmental science gather. Find a way of processing more data and new research doors will be thrown open.

"I believe grid computing will revolutionise the way we compute, in much the same way as the World Wide Web and internet changed the way we communicate," says Dr John Ellis, a theoretical physicist and adviser to the director-general of CERN.

READ MORE

The metaphor is appropriate - not least because Geneva-based CERN was where British researcher Tim Berners-Lee created and named the World Wide Web a decade ago. Like the internet and Web, a grid network is a computer collective, gaining its breadth and power from the massed, individual roles played by many machines. Where the internet forms a huge communications network, a computing grid harnesses the actual processing power of all those linked PCs, creating a communal supercomputer.

All those processors can be put to work on the constituent elements of very big computing problems, an approach known as distributed computing. It's the computing equivalent of putting a team of skilled workers together to build a house, rather than leaving one person to manage everything alone.

One of the more intriguing aspects of the European grid is that it is an open source software project, using the open source operating system Linux to drive much of the network. Open source projects mean that code developed for the project is shared, free of charge, with a community of interested programmers.

Linux, used for DataGRID because it is considered extremely stable, was produced by a vast army of developers who freely donated time and effort to its development.

Dr Ellis explains grid computing as an explosive moment in the development of computing brainpower, equivalent to the point when mammals leapt ahead of the dinosaurs - "an evolution of electronic intellect and neural capacity, which is a bit like that pre-Cambrian explosion in intelligence".

What's driving DataGRID is CERN's build-out of its "large hadron collider" - a colossal atom-smashing facility that will form a 27-kilometre, four-storey high underground ring just outside Geneva. In contrast, existing colliders such as Stanford University's linear accelerator are three kilometres in length. The hadron collider is expected to generate several petabytes of data annually - more than any existing supercomputer or grid can cope with.

A petabyte is a barely imaginable amount of information - a thousand million million bytes (or a quadrillion, in US terminology). This is about the same size as all the information that could be held on 125,000 PC hard drives. IBM explains the figure as enough storage capacity to hold all the information contained in half of all the academic research libraries in the US, or equivalent to printed pages produced from 50 million trees.

Researchers need DataGRID to be online by 2006, when the hadron collider is scheduled for completion. Thus, the largely EU-funded DataGRID "has a very clear job description and a very clear timescale", says Dr Ellis.

Delays are always a worry, he acknowledges, but CERN has interim deadlines by which the grid must process specified sets of data. That is helping to keep the project focused.

But the project is not just of benefit to CERN. While DataGRID will be centred and managed at CERN, many other nodes - webs of computers that form the overall grid - will be spread across Europe. European astronomers will probably use DataGRID, as will scientists studying global weather patterns, biologists crunching data from genetics research and more. Organisations such as the European Space Agency are involved, as are many national science agencies across Europe. IBM is also a member of the consortium.

Several large grid projects are under way in the US and DataGRID will become an equal partner with those. The Republic's grid, known as GRIDIreland (www.grid-ireland.org), is informally affiliated with DataGRID, says Dr Brian Coghlan of TCD. Begun in 1999 and currently linking Trinity, UCC, UCG and Queen's University Belfast, researchers hope to expand GRIDIreland to include nine sites, including the Dublin Institute of Advanced Studies. Dr Coghlan says TCD will be part of the test-bed for DataGRID applications.

Despite such Irish involvement, the State remains the only European nation along with Luxembourg not to be a full member of CERN. This is a sore point for many Irish researchers, who feel the Republic needs to be a full partner in European scientific endeavour, not a watcher from the sidelines.

However, an advisory committee from Georgia Tech, working on behalf of Forfβs, recently recommended that the State not join CERN. It argued that the State can avail of much of CERN's research anyway and the membership cost is high, at around 9.5 million Swiss francs (€5 million) annually.

Dr Ellis notes the State's commitment to science research and development has been low historically, because of its agrarian background and a lack of funds. But he argues this has changed, with the State recognised as a centre for high technology.

"Some EU partners regard it as somewhat of an anomaly that a very rich country is not somehow involved," he says, and many may start to ask: "What are you contributing to the pot?" Membership allows full access to "a network of institutions, facilities and people".

To date, many Irish researchers have had to rely on the benevolence of personal contacts within CERN to participate in research projects, says one academic source.

"There's a general feeling that now is the time to reinforce the culture of science in Ireland," Dr Ellis says - not least because research is often associated "with future wealth creation".