Clustered indexing technique for multidimensional index structures

Abstract

This paper presents an index clustering technique called the segmented page indexing (SP-indexing) for multidimensional index structures. The design objectives of the SP-indexing are twofold: (1) to improve the range query performance of the multidimensional indexing methods and (2) to provide a compromise between optimal index clustering and excessive full index reorganization overhead. The SP-indexing uses two kinds of I/O units: pages for random disk accesses and segments for sequential accesses. The SPindexing improves the range query performance by offering high-performance sequential disk access within a segment. Experimental results demonstrate that the SP-indexing improves the range query performance up to several times compared with the traditional page-based indexing methods with respect to thetotal elapsed time. © Springer-Verlag Berlin Heidelberg 2002.