Efficient Strategies for Workload Distribution in Heterogeneous Computing Systems

An effective workload distribution has a prime rule on reducing the total execution time of a parallel application on heterogeneous environments, such as computational grids and heterogeneous clusters. This paper presents two approaches to workload distribution based on analytical models developed to performance prediction of parallel applications, named VRP and VRP-SS. The workload is distributed based on relative performance ratios, obtained by analytical models constructed according to the rules described in the PEMPIs methodology. In the experimental tests we evaluated the strategies and compared them using MPI applications. The results showed the efficiency of our strategies by reducing the total execution time and improving the workload balance.