DiaGrid (distributed computing network)

DiaGrid (distributed computing network)

DiaGrid is a large high-throughput distributed research computing network utilizing the Condor system and centered at Purdue University in West Lafayette, Indiana. DiaGrid received a 2009 Campus Technology Innovators award from Campus Technology magazine[1] and was employed at the SC09 supercomputing conference in Portland, Ore., to capture nearly 150 days of compute time for science jobs.[2]

The grid, a partnership with Indiana University, Indiana State University, the University of Notre Dame, the University of Louisville, the University of Wisconsin, Purdue's Calumet and North Central campuses, and Indiana University-Purdue University Fort Wayne, is designed to accommodate computers at other campuses across the nation as new members join. In 2009, it included nearly 30,000 processors. The Purdue portion of the pool, named BoilerGrid, is the largest academic system of its kind. In 2008, users tapped DiaGrid’s 177 teraflops of computing power for 16 million compute hours, up from 10 million in 2007. Also in 2008, DiaGrid ran nearly 12 million jobs, 2 million more than in 2007. Since 2006, the system has provided nearly 50 million compute hours to more than 300 researchers.

Through Condor, developed at the University of Wisconsin, DiaGrid harvests and manages computing cycles from idle or underused high performance computing cluster nodes, machines in campus computer and other labs and office computers. Whenever a local user or scheduled job needs a given machine, the Condor job is stopped and sent to another Condor node as soon as possible. While this "opportunistic" model limits the ability to do parallel processing and communications, a Condor pool can provide smaller, serial jobs vast numbers of cycles in a very short amount of time. Condor—and by extension, DiaGrid—is designed for high-throughput computing and is excellent for parameter sweeps, Monte Carlo simulation, or nearly any serial application. Some classes of parallel jobs (master-worker) may be run effectively via Condor as well.

DiaGrid is managed by Information Technology at Purdue (ITaP), the central information technology organization at Purdue's West Lafayette campus, which also operates the Steele and Coates supercomputers. To pool computational resources spread around the state, the grid takes advantage of I-Light, the high-speed fiber-optic state network connecting Indiana campuses to each other, the Internet and national research networks such as the Internet2 and National LambdaRail.

DiaGrid and BoilerGrid have been used by researchers at Purdue and elsewhere for a variety of purposes,[1] such as imaging the structure of viruses at near-atomic resolutions,[3][4] simulating the early stages of the Solar System's formation, projecting the reliability of Indiana's electrical supply, modeling the spread of water pollutants, and identifying millions of potential new forms of zeolites, silicate minerals widely used to catalyze chemical reactions on an industrial scale. DiaGrid provides computational resources to researchers on both the Open Science Grid and the TeraGrid.

External links

Citations