In its October 1992 Report, the Vice Chancellor's Advisory Committee on Computing and Systems recommended that the University explore state-of-the-art research computing options that capitalize on new technologies and utilize a distributed computing model. In response to that recommendation, the University has embarked on a demonstration project to explore the development of a carefully configured cluster of high performance workstations serving a clearly defined community of high-end researchers.
Because the numerically intensive computing applications, graphics rendering, and parallel processing research of the CUNY Ph.D. Program in Computer Science will provide a rigorous test of the clustered environment, it has been selected to serve as the basis of the pilot study, and the project has been developed in close consultation with Dr. Stanley Habib, Executive Officer of the Ph.D. Program and the program's council of chairpersons. In addition, the cluster is a joint study of the CUNY Office of Computing and Information Services and the IBM Corporation.
The cluster has been installed at the Graduate Center and became operational in Fall 1993. It is accessible to its community of users via telnet and ftp within the current CUNYNet configuration. Given the profile of parallel applications now under development within the University, a seven node cluster of IBM RISC System/6000 Models 340 providing computer service, with IBM RISC System/6000 Models 350 providing file and terminal service, has been configured and networked with both 10Mb Ethernet and 100Mb FDDI.
New and emerging software technologies that provide for parallel and load-balancing capabilities are central to this effort. CUNY's Graduate Computer Science faculty have proven expertise in products such as Scientific Associates' Network LINDA and the University of Tennessee's PVM. Sophisticated batch and load balancing software (e.g., Sterling's NQS/Exec) will add diversity to the environment, and a shared file system across the Ph.D. Computer Science program will be designed to provide students and faculty alike with valuable experience in late-breaking technology, and will allow researchers to take full advantage of the computing resources at hand.
Research projects already identified include work in parallel algorithms, graphics and scientific visualization, medical informatics, artificial intelligence, natural language processing, machine architectures, network topologies and optimization.
-- Colette A. Wagner