TACC, IBM UT Grid Project Celebrates Successful First Year

The ambitious UT Grid Project, a joint project between the Texas Advanced Computing Center (TACC) at The University of Texas at Austin and IBM Corporation, marked its first year with the introduction of two new grid computing resources for computational research. Working in collaboration with Austin-based United Devices and the Condor Project at the University of Wisconsin, TACC and IBM have deployed two high-throughput computing resources – Roundup and Rodeo. Both resources harness the idle cycles of PCs for use by researchers who need very large numbers of cycles on problems that can be broken apart into many smaller, independent tasks. “We’re excited to offer these new computational resources that will improve research and education at our university,” said Maytal Dahan, TACC project manager for UT Grid. “Roundup and Rodeo offer very different capabilities from our traditional parallel computing platforms, especially as they grow to include 10,000 or more PCs across campus.” The Roundup user environment currently includes about 1,000 PCs and laptops located on the university campus. Some users have scheduled large jobs that harvest the free cycles and use the array as a single resource. Roundup relies on software from United Devices. The other environment, Rodeo, is a set of pools or "flocks" of PCs and workstations configured within the Condor resource management system. Rodeo has a dedicated central management system at TACC, which can flock to Condor pools at the university's Computer Sciences department and at the Institute for Computational and Engineering Sciences. Several hundred nodes are now in production across the three pools. “This is one of the first steps in enhancing the capabilities of researchers and educators on the university campus via grid computing technologies,” said Dr. Jay Boisseau, director of TACC and the UT Grid project. “Over the next year, we will enhance Roundup and Rodeo and provide new capabilities for campus wide use of storage, visualization and data resources.” The strategy for building UT Grid is to integrate diverse campus resources, from personal-scale to terascale, into a campus wide cyber infrastructure available to the entire university community. With a common security infrastructure and grid computing middleware installed on campus resources, UT Grid will enable higher performance and advanced capabilities in computational research, new paradigms for educational applications, and innovation in grid computing itself. The first year of the project also resulted in the development of GridShell, an innovative software program that makes the use of diverse, distributed resources easier by transparently incorporating grid computing concepts into the UNIX shell login environment. The initial version of GridShell is in beta and for Linux only, but it is scheduled to be released for production use this summer. GridShell allows users to integrate their own PC directly into UT Grid making the use of UT Grid resources almost transparent. Dr. Edward Walker, chief technologist for UT Grid, said, “The goal of Grid computing is to enhance a user’s productivity and not complicate it with new technology. GridShell hides this complexity by transparently assisting users in doing their work on the Grid.” Among the UT Grid’s goals in year two are to package and document Gridshell for widespread use and to complete a new version of TACC’s GridPort software. Accomplishing these objectives will allow TACC to build strong links among computational resources (PCs and laptops, servers, clusters); storage resources (local hard disks, storage area networks, tape archives); visualization equipment (PCs, workstations, special-purpose display equipment, projection rooms); data collections (stored in diverse database management systems); and instruments (large-scale telescopes , electron microscopes, distributed sensor arrays). Much of this work is already in progress, both by UT Grid developers and by leveraging work of the National Science Foundation (NSF) TeraGrid, of which TACC is a member. “TACC is committed to producing software technologies that enhance people’s ability to conduct research, teach and learn, and this project is our exemplar of this commitment,” Boisseau said. “We are eager to help demonstrate the value of grid computing technologies in research and education through the activities of UT Grid.” Background: The Texas Advanced Computing Center (TACC) at The University of Texas at Austin is a leading research center for advanced computational science, engineering and technology. TACC’s mission is to advance the research and education programs of the university and its partners through the application of advanced computing technologies. To fulfill this mission, TACC provides comprehensive advanced computing resources and support services to researchers in Texas and across the nation. TACC also conducts research and development in applications and algorithms, in computing systems design/architecture, and in programming tools and environments to produce new technologies that expand the capabilities of researchers for knowledge discovery. Please visit its Web site. The UT Grid Project is a comprehensive cyber infrastructure project in collaboration with IBM Corporation to create a distributed knowledge environment at The University of Texas at Austin for research and education. The goal of UT Grid is to enable UT researchers to make significant advances in scientific discovery through campus wide access to computing, visualization, data, and instrument resources, as well as to facilitate new educational models based on this infrastructure. UT Grid will also serve as a model campus grid connecting all resources at The University of Texas at Austin, from the high-end computing systems at TACC to departmental clusters to the desktops and laptops of faculty, staff and students. Please visit the projectWeb site.