Grid networks and high-performance computing clusters are increasingly being used in scientific research to carry out large-scale number crunching in place of traditional supercomputers. Some grid computing systems have also found applications in business, particularly in financial services.
Oxford's computing cluster, made up of 10 Dell Poweredge 2650 servers, analyses proton research data from the Fermi National Accelerator Laboratory in Illinois. It investigates mass, inertia and antimatter and studies the behaviour and effects of elementary particles on the universe.
Todd Huffman, project leader at Oxford University, said, "Clustering offers the power of a traditional proprietary supercomputer but at a fraction of the implementation and management cost." The university chose clustering over a grid model because it wanted to keep close track of its data and how the processing power is used, he added.
Another advantage of clustering is that the university will be able to expand and upgrade its infrastructure by adding processors and storage capacity as needs arise.
Oxford researchers will collaborate with teams at Liverpool University, University College London and Glasgow University, which also have computing clusters based on Dell hardware. In future, they will look to integrate the Oxford cluster more closely with those at the other universities and incorporate a batch of 16 dual-processor IBM machines located on another site.
"The ultimate goal is to figure out why there is matter in the universe and not equal parts of matter and antimatter," Huffman explained.