Warren Smith, a nationally-known Grid computing expert, has joined the Texas Advanced Computing Center (TACC) at The University of Texas as manager of the Distributed and Grid Computing (DGC) group. Formerly with the Information Power Grid (IPG) project at the NASA Ames Research Center, Smith will start at TACC in mid-January.
“I'm very excited to be joining TACC, which is a rising star in the high-performance computing world,” Smith said. “TACC is poised to make major contributions to Grid computing, and I think it will be great to be a part of it. One of the main challenges to address is providing a software environment that allows users to take advantage of resources that vary from the worlds fastest supercomputers and visualization devices down to the desktops and laptops of professors and students. The TeraGrid, TIGRE, and UT Grid projects that TACC participates in include all of these resources. Putting all that together will be a fascinating challenge.”
“We are excited to have a researcher of Warren's ability and experience joining our team to lead our distributed and Grid computing activities,” said Jay Boisseau, director of TACC. “TACC already has several prominent projects?TeraGrid, UT Grid, TIGRE, GridPort, and GridShell?and with Warren's leadership TACC will continue its growth towards becoming a world leader in Grid computing technology R&D and deployment.”
Smith noted that Grid computing is justly perceived as a world of difficult problems. “Grids are complex by their very nature,” he said, “uniting many participants and many resources with a broad range of activities and goals. One challenge is investigating the many research problems that arise when trying to construct a Grid environment. Another challenge is taking this research and providing a reliable and easy-to-use software environment that is constructed from many pieces of software.”
Smith has worked on both ends of the Grid problem. He has written articles on the theories of scheduling, information management and Grid management, “and I've also had hands-on experience with an operational Grid system, the IPG at NASA, so I've done some real-world Grid-wrangling,” he said.
Smith graduated from The Johns Hopkins University with a B.S. in electrical and computer engineering and an M.S. in computer science, awarded concurrently in 1992. Smith obtained an M.S. in electrical and computer engineering from Northwestern University in 1995 and his doctorate in electrical and computer engineering from Northwestern University in December 1999.
His dissertation, “Resource Management in Metacomputing Systems,” was directed by Professor Valerie E. Taylor, who is now head of the computer science department at Texas A&M University, and Professor Ian Foster of the University of Chicago and Argonne National Laboratory. At Argonne, Smith worked on the original I-WAY project that linked 18 supercomputer sites at SC95 in San Diego, on the Globus project that evolved from the I-WAY, and on techniques for performance prediction, queuing, scheduling, and advance reservations.
At NASA, Smith improved the performance of Grid scheduling algorithms, predicted application performance on the NASA IPG using “instance-based” learning techniques, and developed methods to store and retrieve information in distributed environments. He also architected and led the development of Grid services to aid in the accurate execution of workflows on distributed platforms.
“A number of services, ranging from monitoring and managing the distributed systems to acquiring data, simulating, and visualizing results, need to be available to make Grid computing an easy and effective way to accomplish the goals of scientists,” Smith said. “TACC's GridPort and GridShell projects are unique approaches to the unification of these services that we will be working to perfect.”
As a national leader in Grid computing, Smith served on program committees for major meetings, including the IEEE High Performance Distributed Computing Symposium and the ACM/IEEE Supercomputing conferences. He has also been a member of review panels for large-scale DOE and NSF advanced computation proposals.