Magnetic tape is no longer cost competitive with magnetic disk for storing data. However, for magnetic disk to replace magnetic tape in tape's traditional role of long term, offline data storage, a new innovative data storage architecture and system must be developed. For DOE's nuclear physics facilities that generate several hundred terabytes of data annually, and that are projected to generate petabytes of data in the near future, switching from magnetic tape to magnetic disk for long term data storage would save a large amount of money by moving from expensive lower volume tape technology to commodity off the shelf disk technology. This project will build and extensively test a magnetic disk storage system using commodity off the shelf components and develop the necessary system software to use the magnetic disks in the way magnetic tapes are used, i.e., Disk AS Tape (DAST). The system will feature the characteristics of a typical magnetic tape jukebox in terms of power consumption, cooling requirements, transfer rate, and operations; but will use less physical space and offer tremendous scalability. In Phase I, a 15 terabyte magnetic disk storage system will be built, the operating system will be modified to keep the disk drives powered down when not in use and to support thousands of disk drives, and the I/O capacity of the system will be measured and tested both locally and overall multiple gigabit ethernet interfaces. Test data will be written to the disk drives in several industry standard formats and the disk drives will be transferred to standard workstations to verify interoperatability and to determine the best format for DAST.
Commercial Applications and Other Benefits as described by the awardee: Although several small DAST implementations have already been deployed for commercial data backup, the proposed research should allow DAST to be scaled to large deployments and more complete solutions. Initially, four markets will be targeted: (1) commercial data backup and data archive, (2) geophysical seismic data storage, (3) remote imaging storage, and (4) scientific data storage including high energy and nuclear physics. Overall, products based on this technology should take a significant share of the multibillion dollar magnetic tape market. Beyond the magnetic tape market, there is tremendous explosion of data in both govenment and industry. Much of this data is stored only temporarily and then discarded even though it has potential long term value because existing large scale data storage systems are too expensive
