Heuristic model for task allocation in distributed computer systems

In distributed computing systems, partitioning of application software into modules and proper allocation of modules among processors are important factors for efficient utilisation of resources. A method for static allocation of modules to processors, with the constraints of minimising interprocessor communication cost and load balancing is presented. The heuristic approach forms module clusters around maximally linked modules or attached modules, and restricts the cluster size to the average load to be assigned to each processor. While modules are being allocated, specific capabilities of the processors can also be taken into consideration. The module allocation with the above constraints is carried out in a single phase, and the algorithm provides a near optimal solution.