Towards a communication-aware task scheduling strategy for heterogeneous systems

Many research activities have focused on the problem of task scheduling in heterogeneous systems from the computational point of view. However, a sche duling strategy should also take into account the communication requirement s of the applications and the communication bandwidth offered by the networ k. Towards this end, in this paper we first propose a model of communicatio n cost between network nodes. This model can be used to properly characteri ze the existing network resources. Second, we propose a criterion to measur e the suitability of each allocation of network resources to each parallel application, according to the communication requirements. Third, we propose a scheduling technique based exclusively on this criterion that provides a near-optimal mapping of processes to processors according to the communica tion requirements. Evaluation results show that the use of this scheduling technique fully exploits the available network bandwidth, greatly improving network performance. Therefore, the proposed scheduling technique can be u sed in the design of communication-aware scheduling strategies for those si tuations where the communication requirements are the system performance bo ttleneck.