Power players in the quest for more potent computing
Analyzing billions of molecules for their cancer-fighting potential is a daunting task. So is the work of biochemists, computational biologists, and life and genetic scientists who examine millions of compounds, strands of DNA and structures to understand the cause, treatment and prevention of genetic diseases.
Enter grid computing, a technology model that harnesses the power of computers located throughout the world. Unlimited by physical proximity and operating heterogeneity, grid computing works to provide extremely potent computing horsepower on demand.
The University of Texas at Arlington is emerging as a leader in this rapidly developing field.
“Grid computing is one of the newest ideas in computing,” said physics Professor Kaushik De, who recently led the University’s participation in an experiment linking tens of thousands of computers to create one of the world’s largest data grids. “It is a natural evolution of the World Wide Web where not only information but computing resources are shared.”
Grid computing leads to lower costs, better collaboration through virtual organizations or dynamically changing groups, and increased flexibility and resource options for large-scale efforts. It handles tasks that are too big or too computing-cycle intensive for individual computers or systems.
It provides access to vast amounts of data in distributed databases or file systems. Furthermore, it enables the sharing of highly valued resources and affords new levels of collaboration among organizations and their constituents. Just as an Internet user views a unified instance of content via the Web, a grid user interfaces with a single, powerful virtual computer.
In addition to the work of Dr. De and the High Energy Physics group, UT Arlington’s grid computing résumé includes the Distributed and Parallel Computing Cluster. Headed by computer science and engineering Professor Sharma Chakravarthy, principal investigator for the cluster, the project is a collaboration among UT Arlington’s CSE and Physics departments and the Dermatology Group at The University of Texas Southwestern Medical Center at Dallas.
The DPCC@UTA project received a three-year, $1.4 million National Science Foundation research instrumentation grant. Phase I created a distributed-memory cluster system with more than 200 processors linked to hundreds of terabytes (1 terabyte = 1,000 gigabytes) of storage. Phase II has added three shared-memory multiprocessor systems, each with eight processors and 32 gigabytes of memory, and several terabytes of additional storage.
Dr. Chakravarthy says these systems will help the research groups solve many computer science problems—especially in the area of data mining—that can’t be addressed on individual or even a small number of linked machines.
“I believe this project begins to establish UT Arlington as a national-level, high-performance computing center,” he said. “We plan to seek additional funding over the next five years to add to the facility’s computing power and become part of a national computing grid.”
A third campus project utilizing grid computing is the BioGrid Texas Virtual Research Park and Healthcare Collaborative Network. Paul Medley, assistant dean in the UT Arlington College of Science, and Alexander Zekulin, a senior solutions specialist for IBM, jointly created BioGrid Texas.
Combining engineering and science expertise found at UT Arlington and UT Southwestern Medical Center at Dallas with IBM technology, they established a means to solve large-scale, interdisciplinary science and medical problems.
“We’re bringing incredibly bright people together to do very complex things,” Dr. Medley said. “With BioGrid, we’ve created a virtual, collaborative, problem-solving environment.”
— Avinash Rai