By David McNabb

OIT has reached several milestones toward the creation of a grid computing environment and a high-performance computing facility, as outlined in the University of Maryland’s information technology master plan (discussed by CIO Jeff Huskamp in the Winter 2005 ITforUM). OIT has established an operational Condor-on-WAM pool and integrated it into the university grid infrastructure Lattice Project. Testing of a prototype high-performance computing cluster has also begun.

To quickly but significantly increase computing power available to university researchers, OIT used the Condor System (www.cs.wisc.edu/condor), originally developed at the University of Wisconsin, to identify and utilize otherwise idle computational power in WAM lab computers. Today more than 70 Apple G5s in WAM labs are united via the Condor-on-WAM pool and offer their idle cycles for high-throughput computations. The new Condor pool is currently undergoing testing, and will soon be available for both faculty and student projects. Since Condor scavenges only idle cycles, there is no impact on those using the WAM lab machines for regular activities; as soon as someone logs in at a machine in a lab, any Condor activity is interrupted and completed elsewhere.

OIT has also integrated the Condor-on-WAM resources into the Lattice Project’s grid computing environment (http://lattice.umiacs.umd.edu), begun by Michael P. Cummings of the university’s Center for Bioinformatics and Computational Biology. By expanding on this project, OIT seeks to create a true grid infrastructure that supports distributed computing on geographically and administratively disparate resources.

In addition, OIT is also testing its first prototype high-performance computing cluster. Job control mechanisms, allocation request procedures, and polices for allocation of cycles on this cluster are under development. OIT hopes to make this new resource available for faculty and staff this summer.