GUSTO is a prototype for future computational grids that will link supercomputers, scientific instruments, virtual reality environments, and data archives transparently. It uses software developed by the Globus Project, a multi-institutional collaborative project centered at ANL and ISI. Foster and Kesselman previously won the Global Information Infrastructure Next Generation Award for GUSTO.

One of the award-winning demonstrations on the GUSTO testbed was a real-time, collaborative analysis of data from a microtomographic beamline at ANL's Advanced Photon Source (APS). At the beamline, a team led by Derk Mancini and Ian McNulty used high-brilliance X-rays to probe the 3D structure of a tiny micromachinery part. Data was collected in real time from the APS, reconstructed on an SGI Origin2000 supercomputer at ANL's Mathematics and Computer Science Division, then displayed both at the APS and on an ImmersaDesk virtual reality environment at the convention site. High-speed networks allowed scientists at the two sites to discuss the data.

"The results were stunning," says Foster. "Normally, beamline users travel long distances to use a beamline and then have to wait hours or days to analyze their data. We were showing reconstructed images remotely just 10 minutes after data collection started. The implications for more effective use of this unique scientific instrument are tremendous."

Other applications showcased on the GUSTO testbed included high-throughput computing for quantum Monte Carlo calculations and a record-setting distributed interactive simulation using multiple supercomputers. Scientists at Caltech and the University of Illinois at Urbana-Champaign participated in these applications, and numerous other organizations contributed computer time.

"Collaboration is the key to our success," emphasizes Foster. As evidence, he notes that three of the other eight finalists in the HPC Challenge competition also used Globus components.

Globus research is supported by DARPA, DOE, NASA, and NSF. Work on real-time analysis of APS data is supported by DOE's Mathematical, Information, and Computational Sciences Grand Challenge Program.

For more information on the SC98 HPC Challenge, see http://www.supercomp.org/sc98/hpcc/. For more information on GUSTO and Globus, see http://www.globus.org/.