HOBI: Hierarchically organized bitmap index for indexing dimensional data

Abstract

In this paper we propose a hierarchically organized bitmap index (HOBI) for optimizing star queries that filter data and compute aggregates along a dimension hierarchy. HOBI is created on a dimension hierarchy. The index is composed of hierarchically organized bitmap indexes, one bitmap index for one dimension level. It supports range predicates on dimensional values as well as roll-up operations along a dimension hierarchy. HOBI was implemented on top on Oracle10g and evaluated experimentally. Its performance was compared to a native Oracle bitmap join index. Experiments were run on a real dataset, coming from the biggest East-European Internet auction platform Allegro.pl. The experiments show that HOBI offers better star query performance than the native Oracle bitmap join index. © 2009 Springer Berlin Heidelberg.