Random RAIDs with selective exploitation of redundancy for high performance video servers

This paper jointly addresses the issues of load balancing, fault tolerance, responsiveness, agility, streaming capacity and cost-effectiveness of high-performance storage servers for data-streaming appli-cations such as video-on-demand. Striping the data of each movie across disks in a %andom ” order bal-ances the load while breaking any correlation between user requests and the access pattern to disks. Parity groups are of fixed-size, comprising consecutive blocks of a movie and a derived pari ty block, and resulting in?andom ” disk-members of any given group. Conse-quently, the load of masking a faulty disk is shared by all disk drives, minimizing the degradation in stream-ing capacity. By using the redundant information to avoid accessing an overloaded disk drive, the occa-sional transient imbalance in disk load due t o the ran-domization is partly prevented and, when occurring, can be circumvented. Finally and most important, making a distinction between data blocks and redun-dant blocks and using redundant blocks only when nec-essary is shown to substantially reduce required buffer sizes without giving up the benefits. The result is a simple, flexible and robust video-server architecture.