Particle physicists in the United Kingdom have built the country's largest Grid as part of the GridPP project. Developed to analyse data from the upcoming Large Hadron Collider (LHC) at CERN, the Grid currently spans 17 UK sites and includes almost 10,000 CPUs and more than 500 TB of data storage, with available storage reserves on disk and tape of 1 PB. It is also an integral part of the Enabling Grids for E-sciencE (EGEE) project, contributing most of the computing power for the UK and Ireland.
GridPP is a collaboration of all the UK universities with particle physics departments. It also includes Rutherford Appleton Laboratory (RAL) in Oxfordshire, which acts as the UK Tier-1 centre for LHC computing. Steve Lloyd, GridPP's chair from Queen Mary, University of London, explains the structure: "Apart from the Tier-1, the GridPP sites are joined into four regional Tier-2s: London, SouthGrid, NorthGrid and ScotGrid. These are used to share expertise and support. This works well, particularly for smaller sites where it would be difficult to provide dedicated help on the whole range of Grid issues."
Set up in 2001 initially for three years, GridPP was extended in 2004 to take it up to the switch-on of the LHC in 2007. The project is now funded by the recently created Science and Technology Facilities Council, and a further £30 million was announced last month to extend its remit until 2011.
"The first part of the project aimed to create a prototype Grid – which we did very well," said Dave Britton from Imperial College London, who will be project leader for GridPP3. "We're now focused on running a large-scale, stable, easy-to-use Grid integrated with EGEE and the WLCG [Worldwide Large Hadron Collider Computing Grid]. From later this year we'll see how successful this has been, when we start dealing with the petabytes of data from the LHC."
In addition to providing hardware, GridPP plays a role in applications and middleware. Among other achievements GridPP members have helped develop Grid accounting for the WLCG and EGEE projects, and a user interface for the ATLAS and LHCb physics experiments at LHC. The Real Time Monitor, another GridPP-developed tool, has held centre stage at numerous talks and conferences worldwide. RAL also hosts a Grid Operations Centre that monitors the overall EGEE grid, which has about 200 sites, tens of thousands of CPUs and petabytes of storage.
GridPP sites have contributed to what is now the EGEE Grid since its very first days, when RAL was one of only five sites in the first European DataGrid test-bed. Although the EGEE Grid now consists of hundreds of sites, the UK still provides a substantial proportion of its infrastructure – in the recent biomedical data challenge searching for drugs against malaria, known as WISDOM, GridPP provided nearly half the computing hours used.
Working with non-particle physicists is an important part of GridPP’s remit. "Although our Grid was built to analyse particle physics data, when we have spare capacity we're able to share it with other scientists worldwide," said Tony Doyle, GridPP’s current project leader. "We're happy to have contributed millions of hours of computer time to help find drugs against malaria and avian flu."
More information is available on the project’s website (www.gridpp.ac.uk>).
• This article was published in iSGTW on 11 April.