Hierarchical declustering schemes for range queries

Abstract

Declustering schemes have been widely used to speed up access time of multi-device storage systems (e.g. disk arrays) in modern geospatial applications. A declustering scheme distributes data items among multiple devices, thus enabling parallel I/O access and speeding up query response time. To date, effcient declustering schemes are only known for small values of M, where M is the number of devices. For large values of M, the search overhead to find an efficient scheme is prohibitive. In this paper, we present an efficient hierarchical technique for building declustering schemes for large values of M based on declustering schemes for small values of M. Using this technique, one may easily construct efficient declustering schemes for large values of M using any known good declustering schemes for small values of M. We analyze the performance of the declustering schemes generated by our technique in 2-dimension, giving tight asymptotic bounds on their response time. For example we show, in 2 dimension, that using optimal declustering schemes for M1 and M2 devices we can construct a scheme for M1M2 devices whose response time is at most seven more than the optimal response time. Our technique generalizes to any d dimension. We also present simulation results to evaluate the performance of our scheme in practice.