We have developed two techniques that exploit the characteristics of the data being accessed to reduce the failure recovery overhead. The first technique exploits the sequentiality of video data access to reduce the on-line failure recovery overhead in parity-based disk arrays. We have shown that this technique reduces the worst-case failure recovery overhead from 100% to 1/(G-1) in RAID-5 arrays, where G is the parity group size; and from (G-1)/(C-1) to 1/(C-1) in de-clustered parity arrays, where C is the cluster size. The second technique utilizes the inherent redundancy in video streams (rather than error-correcting codes) to ensure that the on-line failure recovery process does not impose any additional load on the disk array. We have developed a disk array architecture that enhances the scalability of multimedia servers by: (1) integrating the recovery process with the decompression of video streams, and thereby distributing the reconstruction process across the clients; and (2) supporting graceful degradation in the quality of recovered images with increase in the number of disk failures.

Representative Publications:

1. P. Shenoy and H.M. Vin, Failure Recovery Algorithms for Multimedia Servers, ACM Multimedia Systems Journal, Vol. 8, No. 1, pp. 1-19, January 2000. [ Abstract | Paper ]

2. P. Shenoy and H.M. Vin, Efficient Support for Scan Operations in Multimedia Servers, In Proceedings of the ACM Multimedia'95, San Francisco, CA, pp. 131-140, November 1995 [ Abstract | Paper ]