In 1984 data associated with the Atlas Centre's computers was held either on disk or in a library of about 60,000 manually-handled reels of magnetic tape. Data volumes were continuing to grow rapidly and there was a pressing need to find better and less labour-intensive ways of coping with them and of making them accessible from services other than the IBM mainframe.
By 1994 the concept of an Atlas Data Store had been developed and implemented. This was a filestore designed to handle automatically multi-Terabytes of data composed of large numbers of multi-Megabyte files, using a hierarchy of storage media, and accessible conveniently over local and wide-area networks speedily and without using unfamiliar commands. The design was such that the actual storage devices and media could evolve with time to take advantage of whichever were most cost-effective and appropriate for the volumes of data being handled.
This was a substantial project with many interim steps on the way and involved a succession of robotic systems (Masstor M860, StorageTek 4400 and IBM 3494) for housing and retrieving data. More significantly, the project required major software developments to provide the required functionalities and performance, and to keep abreast of the constantly evolving environment of computer systems and network architectures with which the data store had to interact.
