SU joins New York State higher education grid computing project

SU joins New York State higher education grid computing projectSeptember 13, 2006Judy Holmesjlholmes@syr.edu

Syracuse University has joined forces with seven public and private universities in a new venture to provide researchers access to high-powered computing resources through the New York State Grid. In addition to SU, the universities involved in the project are the University at Buffalo (project lead), Binghamton University, the University at Albany, Cornell University, New York University, the Rochester Institute of Technology and the University of Rochester.

In order to join the Grid, SU is building a prototype Beowulf Class computer cluster, which will eventually be available to all researchers on campus and centrally supported by SU’s Information Technology and Services (ITS). To learn more about the new SU computer cluster and the University’s role in the New York State Grid project, all SU researchers are invited to attend a luncheon at noon on Sept. 20, hosted by Ben Ware, vice president for research and dean of the Graduate School. Please RSVP via e-mail to Kathleen Leavens at klleaven@syr.edu by no later than Sept. 18.

“Sharing computing resources expands our capabilities and reduces our costs,” Ware says. “The NYS Grid is the best opportunity we have seen to participate in a generalized grid development that can benefit researchers in many different fields of research.”

The centrally supported Beowulf computer cluster will be the third “super computer” cluster available at SU. Five years ago, the Department of Physics in The College of Arts and Sciences built the Weasel, a second-generation, 102-node Beowulf cluster; three years ago, the Maxwell School built a 37-node Beowulf cluster. Both clusters are working at maximum capacity.

The ITS-supported cluster will initially have a much smaller local capacity than either the Weasel or the Maxwell School cluster, however, the ITS cluster will be the gateway to the NYS Grid, which could provide virtually unlimited computing resources for the SU research community.

“Scientists will be able to log into the grid, indicate the amount of CPU resources and computing time they need, and the grid software will look at all available resources and grant access to a cluster that will enable them to do their calculations,” says Jorge Gonzalez-Outeiri?o, scientific IT analyst for SU’s Information Technology and Services.

Gonzalez-Outeiri?o, who has a Ph.D. in chemistry, is building the SU cluster and is the SU representative on the NYS Grid project. Before arriving at SU in April 2005, Gonzalez-Outeiri?o was a postdoctoral research associate in the Complex Carbohydrate Research Center at the University of Georgia, where his research group, led by Robert J. Woods, built the center’s first Beowulf cluster. “We called it the eBay cluster, because we initially had few funds and bought most of the parts on eBay,” Gonzalez-Outeiri?o says.

However, that eBay cluster, and the later addition of two more Beowulf clusters, enabled Gonzalez-Outeiri?o and his partners to conduct groundbreaking research that is part of the first steps in developing a vaccine for Group B, Type III streptococcus (GBS-III). GBS-III is carried by an estimated 25 percent of pregnant women and is responsible for life-threatening illnesses in newborn babies. Gonzalez-Outeiri?o’s research group was the first to predict a model of the three-dimensional structure of the GBS-III protein-carbohydrate complex and its components. The discoveries required more than a month of computer resources running 24 hours a day, seven days a week. The research has been published in the prestigious Proceedings of the National Academy of Sciences (2006) and in Carbohydrate Research (2005).

“Without the Beowulf computing resources, we would not have been able to do the simulations needed to determine the molecule’s structure,” Gonzalez-Outeiri?o says. “We need to provide these kinds of computing services to all scientists. That is why SU is involved in the grid computing project.”