On Exploitation vs Exploration of Solution Space for Grid Scheduling

Finding an optimal solution is computationally hard for most combinatorial optimization problems. Therefore the use of heuristics methods aims at finding, if not optimal, near optimal solutions in reasonable amount of computation time. Due to lack of knowledge about the landscape of fitness function, searching the solution space by heuristic methods becomes very challenging. One can search the solution space through a path of feasible solutions, in which the search method manages one solution at each iteration. This is known as exploitation of the search space, as it enables going deep into some close area of a solution. One the other hand, one can use heuristic methods that manage many feasible solutions at an iteration, the population-based heuristics. This is known as exploration of solution space as it enables a wide search in the solution space. The question is then, for a given combinatorial optimization problem, which of the two search methods is more effective. In this work we present a study on the effectiveness of using exploitation vs exploration search for the problem of Independent Batch Scheduling in Computational Grids. We also consider the combination of both search processes, which showed to effectively solve the problem.