A Unix-MVS based mass storage system for supercomputers

The design, implementation, and current utilization of a large mass storage system (MSS) is described. The MSS supports a large, networked, multivendor, Unix-based supercomputing facility. The MSS provides all processors on the network with the ability to store large amounts of data in a highly accessible long-term repository. Processors on this network vary in size and capability from personal computers and workstations to multiple large supercomputers like the CRAY-2 and CRAY Y-MP. The MSS emulates a Unix System V file system and is capable of storing hundreds of thousands of files ranging in size from a few bytes to 2 Gbytes. This use of Unix reduces the training requirements and allows the use of existing Unix networking solutions along with locally developed networking enhancements. The systems are currently networked via the NSC HYPERchannel and Ethernet. To assure data integrity and provide volume spanning capabilities, the IBM MVS operating system was integrated into the MSS. The MSS implementation was designed to make MVS transparent to the MSS user.<
>