Laxity threshold polling for scalable real-time/non-real-time scheduling

Minimum Laxity Threshold scheduling has been found by several authors to give a near optimal trade-off between soft-real-time loss rate due to expired deadline and non-real-time average delay. By varying the threshold of the scheduling algorithm the real-time/non-real-time quality of service trade-off may be adjusted. In this paper an approximation is proposed to minimum deadline threshold scheduling. By means of discrete event simulation the approximation is demonstrated to come close to minimum deadline threshold scheduling in its performance. This approximation has the advantage of giving better scalability than minimum laxity threshold. It also has the advantage of not requiring a single queue that must be scheduled so it is usable for distributed applications.