Due to their high storage and bandwidth requirements, multimedia applications place heavy demands on the storage system. The performance of I/O devices (in particular, tapes and disks) lags behind the performance of processors and network systems, resulting in I/O becoming the bottleneck in current systems. This gap is expected to increase in the future since I/O performance is limited by physical motion. In such a scenario, it is imperative that novel techniques for improving I/O performance be developed. Traditional methods such as RAID are not designed for multimedia data and thus do not yield optimal performance. Using knowledge of the application domain, it is possible to achieve much more efficient I/O than more general solution. We are investigating the use of data placement, prefetching and I/O for multimedia applications. The judicious placement of data across multiple disks or I/O nodes is used to overlap latency for multiple requests and deliver data at higher bandwidths. We have developed declustering schemes for multidimensional range and similarity queries. These schemes have been shown, through simulation, to have superior performance to existing schemes. We will investigate their performance through experimentation with real data and real disks, and integration with prefetching and I/O scheduling.
Annual Research Report