It is well known that normally only a fraction of the whole datastore of an organization is being accessed daily. And only a limited amount of data is added to the archive every 24 hours. These simple considerations, and also the fact that it is a way cheaper to store data on tapes than on disks suggest that a combined disk/tape storage solution is the most efficient one. The basic principles to follow are these: "1) provide enough online disk space for daily work, 2) enough tape space to host the whole archive, and 3) means for disk-to-tape migration".

At CASPUR, we also opted for a mixed disk/tape storage architecture. And, after considered both staging and HSM solutions for data migration, we decided that staging fits better our goals.

The first staging solution was put in production at CASPUR in 1997. After more than 8 years of operation, this system made into a quite reasonable product (bugs were fixed, perfomance was tuned, redundancy ensured etc). We believe it has reached the point when it may be shared with the others.

Our system provides a link between an automated tape pool and one or more normal filesystems where users customarily operate. Users gain an option to move files back and forth between disks and tapes in a transparent way. With CASPUR solution, it becomes possible to maintain an individual level of occupancy for each of the file systems involved (long-unused files are being moved automatically to tapes).