IN THIS ARTICLE    »  The Need for a Competitive Edge    »  A Messy Mix of Hardware    »  A Total Solution and Immediate Implementation    »  Providing A Competitive Advantage    »  Key Results and Benefits

The Physics Department of the University of Oxford is one of the largest in the United Kingdom, with an award-winning research program that covers everything from quantum computing to Martian atmospheres. This work attracts both students looking for the best preparation for physics-based careers, as well as substantial funding from research councils, industry, and the European Commission.

As research requirements become more complex, the demands on the department's computing capabilities grow accordingly. With a history of substantial contributions to their field, even producing Nobel prize-winning graduates, the sub-department of Theoretical Physics (TP) especially requires heavy computational analysis for scientific research.

But with constant pressure to stay competitive among the foremost universities, the TP sub-departments' hardware platform was not living up to the job. Oxford turned to Sun Microsystems to provide a total hardware and networking solution.

The Need for a Competitive Edge
Lory Rice, computer manager of the Theoretical Physics sub-department, explains his department's computing needs: "TP's computing needs are three-fold: Heavy computational analysis for scientific research entailing hours, days and even weeks of computing. Second is the preparation of scientific papers for journals, presentations for talks, and the drafts for scientific books. Third is communication between colleagues throughout the world, principally by email but with other computer-based technologies evolving rapidly."

According to Rica, Oxford TP has a fundamental and ongoing need to increase its processing capacity. "The reason that needs have grown is just a reflection of the way that computing needs have changed and grown in the wider society. In the case of theoretical physicists, they can undertake more complex research requiring greater computational power just because appropriate computers are available. Much of today's research, especially in the physical sciences, could not be done before the advent of today's powerful computers."

A Messy Mix of Hardware
When Rice joined Oxford in 1995, he inherited an inefficient and disconnected mixture of hardware. "In the early '90s, the Theoretical Physics sub-department moved away from relying on VT100-compatible terminals connected to VAX computers serving the whole of the Physics Department to its own network of first-generation SPARC servers and processor technology," Rice says. "Driven by the familiar mantra within universities of 'value for money,' by 1995, when I arrived to take charge of TP's computers, this purchasing policy had given rise to a mixture of perhaps 40 machines which included Silicon Graphics Indys and DEC Alphas as well as the SPARCs."

Not only was the system inefficient and disjointed, but Rice saw computing and servicing needs growing beyond what this ad hoc method could handle. "With a growth to an eventual doubling in the number of machines, to give everyone a desktop computer, it was becoming impossible to provide adequate support with three operating systems on this number of machines."

Rice chose to resist this ineffective purchasing policy and started looking for the best hardware to create a streamlined and unified system. He found the complete solution in a combination of Sun Microsystems products.

A Total Solution and Immediate Implementation
Sun Microsystems put forward an unmatched, complete solution, providing the high-quality hardware required to meet the demanding needs of the TP sub-department. The solution included Sun Fire servers and Sun StorEdge SCSI Arrays. Sun brought in Streamline Computing (a specialist in High Performance and Technical Computing and a leading provider of parallel and distributed systems and software) to provide the expertise in the parallel computing software and networking side, including the integration of a high speed interconnect and software stacks.

"The Sun solution, in conjunction with Streamline Computing, offered a cluster of computers organized to run in parallel, thus enabling a large, complex problem to be broken down into smaller parts which could be run simultaneously on separate nodes, enabling the whole calculation to complete more quickly. The new 40-node dual-processor Beowulf Cluster is for research computational analysis and, at present is exclusively used by the TP sub-department."

Sun was the solution for updating the TP desktop needs as well. As Rice explains, "Starting with the UltraSPARC machines in the late '90s, eventually all the machines on the desktop were Sun UltraSPARCs, followed by Ultra 5s and now all Sun Ray ultra-thin clients. Sun also provided central servers for maintaining the network infrastructure."

The implementation went quickly and smoothly, adds Rice. "It was delivered in one day, configured by Sun and Streamline Computing the next and running test codes by our users on the third. It was very quick because it was pre-assembled and tested elsewhere by Sun's distributor and Streamline before delivery to us."

Providing a Competitive Advantage
Thanks to Sun, the theoretical physicists at Oxford now have a streamlined and scalable system, and plans for future expansion. Rice explains, "Calculations never before possible can now be done. Sun has lived up to our expectations. We have a software and hardware support contract with Sun and Streamline Computing and are likely to expand the cluster significantly in the next six months. Without such computers, Oxford TP would not have the tools to compete with other world-class universities in the U.S.A. and elsewhere. Now we have the tools to do so."

For more information on Sun servers, please visit http://www.sun.com/servers

Click here to have your local Sun representative contact you, or email education_news@sun.com