Compressed Datacubes for fast OLAP Applications

  • Authors:
  • Francesco Buccafurri;Domenico Rosaci;Domenico Saccà

  • Affiliations:
  • -;-;-

  • Venue:
  • DaWaK '99 Proceedings of the First International Conference on Data Warehousing and Knowledge Discovery
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

A number of techniques have been proposed in the literature to optimize the querying of datacubes (i.e., matrix representations of multidimensional relations) in OLAP applications. In this paper we are concerned with the problem of providing very fast executions of range queries on datacubes by possibly returning 'approximate' answers. To this end, given a large datacube with non-negative values for the measure attribute, we propose to divide the datacube into blocks of possibly different sizes and to store a number of aggregate data for each of them (number of tuples occurring in the block, the sum of all measure values, minimum and maximum values). Then, when a range query (in particular, count and sum) is issued, we compute the answer on the aggregate data rather than on the actual tuples, thus returning 'approximated' results. We introduce a number of techniques to perform an estimation (with expected value and variance) of range query answers and compare the accuracies of their estimations. We finally present a comparative analysis with other recently proposed techniques; the results confirm the effectiveness of our approach.