Acquisition of CAPRI: CSCC Archival and Parallel Resource Interface

点击翻译按钮获取中文摘要

9512507 Messina The Concurrent Supercomputing Consortium (CSCC) which is led by the California Institute of Technology proposes to acquire a large-scale data storage facility with a parallel I/O architecture and integrate it with the consortium's existing computing resources. It will undertake a CSCC Archive and Parallel Resource Interface (CAPRI) project which will enable new, data-intensive applications to be carried out. The facility will provide the ability to store terabytes of data and to access the storage system with multiple parallel interfaces. It will serve as a key research instrument for the scalable I/O research initiative. The results of the research project will enable the scientists at the Consortium to achieve unprecedented performance on data-intensive computational problems. The project will greatly enhance the Consortium's computing environment by creating storage capability commensurate with the prodigious ability of the computational resources to generate data. The major components of the proposed system are (1) the high-performance storage system (HPSS) software package, (2) a 64 Gigabyte on-board disk cache, (3) 16 HIPPI nodes for supporting parallel network I/O, (4) 80 Gigabytes of RAID disk caches, (5) a 15 Terabyte archival storage system including four IBM 3490-C2A tape drives and robotic tape handling, and (6) a HIPPI switch. ***