Skip to Main Content
Self-scheduling algorithms can achieve a good balance between workload and communication overhead in computational systems. In particular, quadratic self-scheduling (QSS) and exponential self-scheduling (ESS) are flexible enough to adapt to distributed systems. Thus, they are of interest for application in Internet-based grids of computers. However, these algorithms depend on several parameters, which have to be optimized for the working environment. To tackle this problem, we present here a heuristic approach, based in simulated annealing (SA), to optimize all the parameters of QSS and ESS. To such a goal, the computational grid environment is simulated. We find that the optimal SA results permit to reduce the overall computing time of a set of tasks up to a 12%, with respect to results obtained with previous values of the parameters experimentally determined. Moreover, the time to obtain the SA optimized parameters by simulation is negligible compared with that needed using experimental measures. In addition, we find the results to be fairly insensitive to the size of the chunks (sets of tasks sent to a processor). Finally, the results show the SA scheduling approach to be very efficient, since a simple linear dependence of the overall computing time with the number of tasks is found.