Optimization of distributed olap cubes with an adaptive simulated annealing algorithm

Abstract

The materialization of multidimensional structures is a sine qua non condition of performance for OLAP systems. Several proposals have addressed the problem of selecting the optimal set of aggregations for the centralized OLAP approach. But the OLAP structures may also be distributed to capture the known advantages of distributed databases. However, this approach introduces another term into the optimizing equation: space, which generates new inter-node subcubes' dependencies. The problem to solve is the selection of the most appropriate cubes, but also its correct allocation. The optimizing heuristics face now with extra complexity, hardening its searching for solutions. To address this extended problem, this paper proposes a simulated annealing heuristic, which includes an adaptive mechanism, concerning the size of each move of the hill climber. The results of the experimental simulation show that this algorithm is a good solution for this kind of problem, especially when it comes to its remarkable scalability.