BaBar drives new computing technology

UK scientists are to develop and test the next generation of computing technology based upon the massive amounts of data streaming from an international particle physics experiment sited in the USA.

The BaBar experiment at the Stanford Linear Accelerator Centre (SLAC) in California is investigating the nature of B mesons, short-lived sub-atomic particles. But its success in generating data has far outstripped the capability of existing computing to store and process the data.

In order to resolve the situation, the UK’s Particle Physics and Astronomy Research Council (PPARC) and SLAC have appointed the CLRC Rutherford Appleton Laboratory (RAL) in Oxfordshire to develop a ‘shared computer processing network’.

Sharing computer power is not a new idea, but researchers at RAL will take it further than has ever been possible before by employing ‘Grid’ technology. The Grid is an innovative system for networking computers around the globe, allowing them to share the work of storing and processing information, effectively building a worldwide supercomputer.

The aim of the Grid is to allow scientists to log in from their desks to access and manipulate data which may be stored on machines in another country, without needing to go through difficult access protocols. This can be easily understood by comparison with the national electricity grid – users don’t need to know which power station is supplying their electricity, or by what means. They are only concerned with the electricity being available when they need it.

Dr Nicolo de Groot, RAL, says ‘BaBar physicists from the USA, France, Italy, Canada Germany and the UK will be able to access the data through the Grid and get results back on their local computers. Nowadays the computing needs for particle physics are such that we can only meet them if we link facilities across the world and share resources. ‘

By dividing its data storage and processing between several sites, SLAC will be able to take advantage of the new developments coming from the Grid programme, as they become available. BaBar has observed 300 million events to date, which will be stored in the RAL centre. This is expected to double over the next two years.

To understand the data, scientists have to use statistics to simulate an equal number of events. 290 million events have currently been simulated for BaBar, or about 145 Terabytes of information, which if stored on floppy discs, would form a stack 200 miles high. Alternatively, if you lay them side by side, they would form a path from London to Miami – hence the need for new computing technology to handle this vast amount of data.

Professor Roger Barlow, spokesperson for the UK BaBar project said ‘This is really going to move us forward, enabling us to do a better and faster job of understanding our data and recognising what it’s got to tell us about some pretty basic laws of physics.’

Professor Ken Peach, Director of Particle Physics at CLRC, said ‘We are very excited to be given the opportunity to develop this new computer centre for BaBar at RAL. We look forward to the challenge of integrating this into the UK, European and USA prototype Grids and working with our colleagues in the UK and elsewhere to make Grid computing a reality.’

UK and European scientists are currently designing, implementing and testing Grid software, known as middleware, in the EU DataGrid project. The middleware developed and the experience gained will be used to build a Grid for processing the equally enormous amounts of data expected from the Large Hadron Collider (LHC) experiment at CERN, the European Laboratory for Particle Physics.

The EU DataGrid is a 3-year development project due to end in 2003, but UK scientists are welcoming this latest opportunity to test and refine the new systems using real data and processes. The computing centre at RAL, funded by PPARC, will be the first in the UK to seriously test the new technology and will require 300 additional computers this year and over a thousand in the future.

On the web