Proper

Abstract

This paper presents an index-sequential method whose design objectives are twofold: (1) improving the range query performance of multidimensional indexing methods and (2) providing a compromise between optimal index clustering and excessive full index reorganization overhead. Most relevant index nodes (i.e., disk pages) are widely scattered on a disk, and thus many random disk accesses are required during query processing. Our index avoids such scattering by storing the relevant nodes contiguously in a segment that contains a sequence of contiguous disk pages and improves query performance by offering sequential access within a segment. Experimental results demonstrate that our index-sequential method improves query performance up to several times compared with traditional methods.