Implementing High-Performance Grid Computing in Student Computer Labs

摘要

Information technology infrastructure consumes significant resources at most higher education institutions. Innovative ways to more efficiently use this infrastructure are always being sought. Research universities must balance resource consumption between their research and teaching and learning missions.rnThe University of Florida recently installed a grid computing client software product in the University's student computer labs. This product allows faculty with large data analysis research projects to take advantage of the tremendous computing power available through the harvesting of unused CPU capacity available in the University's student computer labs. The grid computing client software runs on the lab computers, and accepts computational jobs from programs submitted to the grid computing system by research users, devoting system resources from each lab PC to the computations whenever they are not being used by the local student user.rnThis presentation will describe the system architecture, costs, implications, and strengths and weaknesses that are important for implementation, including the technical requirements and management system, and will be focused on providing operational staff from other institutions with useful information to help in the evaluation of whether such a system would be effective at that their institution.