Self-Scheduling for a Heterogeneous Distributed Platform

We discuss schedulers for a heterogeneous distributed platform, designed to execute a variety of tasks in a non-dedicated environment. The platform uses and controls a large number of non-dedicated heterogeneous computational resources in a local network. Several self-scheduling algorithms have been adapted to take into account the computational capacity of each workstation of the network. To evaluate the schedulers we use the platform to execute a software tool for molecular docking of a set of 1000 molecules in the 4UDC protein. We analyze the performance of the self-scheduling algorithms and their impact on the execution time of the application.