Spectral Clustering Scheduling Techniques for Tasks With Strict QoS Requirements

Efficient task scheduling is fundamental for the success of the Grids, since it directly affects the Quality of Service (QoS) offered to the users. Efficient scheduling policies should be evaluated based not only on performance metrics that are of interest to the infrastructure side, such as the Grid resources utilization efficiency, but also on user satisfaction metrics, such as the percentage of tasks served by the Grid without violating their QoS requirements. In this paper, the authors propose a scheduling algorithm for tasks with strict timing requirements, given in the form of a desired start and finish time. Their algorithm aims at minimizing the violations of the time constraints, while at the same time minimizing the number of processors used.