Multi-cube computation
Jeffrey Xu Yu, Hongjun Lü · 2001
Computing an n-attribute datacube requires the computation of an aggregate function over all groups generated by 2/sup n/ interrelated GROUP-BYs. In this paper, we focus on multi-cube computation. We extend the algorithms for single datacube computation to process multiple datacubes simultaneously. The issue we intend to explore is the memory utilization. We propose two multi-cube algorithms, namely, a sort-based algorithm and a hash-based algorithm. Different data skews and sparsities are investigated. Results from our extensive performance studies are reported.