On tradeoff between the two compromise factors in assigning tasks on a cluster computing

This paper deals with the static task-assignment problem in a cluster computing system as follows: Given a task composed of a number of interacting modules, assign the task modules to the processors in the system to minimize the communication cost while balancing the processors' loads. Because these two optimization criteria conflict with each other, a compromise needs to be made between them according to the given task type. This paper proposes a new cost function to evaluate the static task assignments and a heuristic algorithm for solving the transformed problem explicitly describing the tradeoff between the two goals. The simulation results showed that this approach outperforms the existing representative approach for a range of tasks and processing systems.