Improved Real-Time Energy Aware Parallel Task Scheduling in a Cluster

Conservation of energy is the main issue in nowadays computing devices. Cluster computing is the new area of research nowadays. This paper deals with two problems in cluster computing, first is scheduling of parallel tasks in a cluster and second is energy conservation. Our main aim is to design a non-preemptive real time Scheduler for Cluster in such a way that overall energy of the processor will be minimized, and meanwhile satisfying the timing constraint of each task. The proposed work is focused on determining the most efficient way to scale down voltages for non-critical tasks. For this purpose we propose two algorithms: one for slack time utilization by non-critical tasks and second algorithm is proposed for duplication of tasks based on some heuristic. These algorithms implement a schedule that provides minimum energy consumption under required conditions.