Analysis needs of rapidly expanding fields of science have pushed the required computational burden to a point beyond the abilities of standard workstations. Novel computational and analytical methods often require many days of computing power, and data storage and integration needs increase with each passing month. To address these needs, we propose to create a shared computational resource comprised of 200 processors (100 2-processor workstations) running the Linux operating system and a storage system with -2 Tb of disk space. These compute nodes will be merged with our existing cluster resources to create a heterogeneous computing environment encompassing approximately 232 processors with a storage capacity of 2.5Tb. This resource can be partitioned to accommodate the needs of a number of individual groups, but also allows the entire resource to be utilized at times of increased need or availability. The resource will be specially targeted to the needs of small- to medium-sized compute-intensive jobs and the development of parallel scientific algorithms, but will also have the required hardward and software resources to allow development of distributed databases, data-warehouses, and parallel database tools for scientific data integration and data mining. Initially, the major user group will be comprised of investigators with primary and secondary appointments from the Department of Human Genetics at the University of Pittsburgh. Additional users will be granted access to the machine on a temporary, proposal-oriented basis, unless they choose to become major users by supporting expansion of the resource.
