Quality of Service Driven Resource Management Algorithms for Network Computing.
Muthucumaru Maheswaran · Parallel and Distributed Processing Techniques and Applications · 1999
This paper presents a novel approach that takes into account the quality of service (QoS) requested by the applications when scheduling the computational resources in a network computing system. The scheduling algorithms examined here are dynamic and centralized. The QoS model used in this study only uses metrics such as application deadlines and application priorities. However, the model is general enough to be extended to handle other QoS metrics. The level of service received by each application is quantified by a benefit function defined for that application. Maximizing the total benefit provided to the applications is the optimization criterion used in designing the dynamic scheduling algorithms. Simulations were performed to evaluate the performance of the algorithms. Some preliminary results from an evaluation study are presented here.