The Future of Mass Storage Systems - Guest Editor's Introduction

the Sixth IEEE Symposium on Mass Storage Systems held at Vail, Colorado, in June 1984. To owners of large collections of data at the symposium, the future looked bright indeed. There were descriptions of new highperformance optical digital data disk products,'-3 descriptions of new very high density magnetic tape4 and magnetic disk formats, and a discussion of the design decisions leading up to the IBM 3480 magnetic tape cartridge.6 To information systems directors, capacity planners, and storage managers, these new announcements meant new choices. They meant alternatives to satisfy the increasing demand for magnetic disks that can consume floor space and power at an alarming rate. They meant alternatives to the slow, error-prone, manual access to magnetic tape libraries. For years, the standard storage paradigm has been a magnetic disk farm serving a cluster of mainframe/main memory units and a manually accessed magnetic tape library for backup and retention of less active files and data sets. There is frequently an off-site archive of magnetic tape and/or microfilm for vital records. The disk farm has also been referred to as a collection of direct access storage devices, or DASD, in the IBM world, and as a secondary store, working store, or local disk elsewhere. Mass storage systems, as opposed to manually accessed tape libraries, have automated the access to a portion of the library. A new physical-volume format is often required, for example, that of the IBM 3850 magnetic tape cartridge. This standard storage paradigm is being altered by a wider need to share stored data and by the rapid growth in storage needs of a hierarchy of performance levels. The high-performance level is the mainframe level with data rates in the 1.5 to 3.0 megabyte-per-second range where the standard storage paradigm applies. The medium-performance range, characterized by specialized workstations and minicomputers with their own local magnetic disks and local area networks in the 0.1 to 1.0 megabyte-per-second range, is expanding. Workstations for CAD/CAM and document management, Ethernet and its competitors, new magnetic disk formats, and optical data disks in the 5.25to 12-inch diameter range and the jukeboxes to serve them are all part of this mid-performance level, and they share rather than occupy mainframe I/0 for access. The low end of the performance hierarchy is characterized by the explosively growing personal computer field. Often, these are clustered with low-speed local area networks. A large part of today's innovation in magnetic, optical, and magneto-optical storage is directed at this lowperformance end of the industry. Albert Hoagland states in his article for this issue of Computer, for example, that the magnetics industry is growing towards a $ 100-billion-a-year industry, with much of the growth expected to be in the