Applying heuristic methods for job scheduling in storage markets

Abstract

In double-sided markets for computing resources an optimal allocation schedule among job offers and requests subject to relevant capacity constraints can be determined. With increasing storage demands and emerging storage services the question how to schedule storage jobs becomes more and more interesting. Since such scheduling problems are often in the class NP-complete an exact computation is not feasible in practice. On the other hand an approximation to the optimal solution can easily be found by means of using heuristics. The problem with this attempt is that the suggested solution may not be exactly optimal and is thus less satisfying. Considering the two above mentioned solution approaches one can clearly find a trade-off between the optimality of the solution and the efficiency to get to a solution at all. This work proposes to apply and combine heuristics in optimization to gain from both of their benefits while reducing the problematic aspects. Following this method it is assumed to get closer to the optimal solution in a shorter time compared to a full optimization.