Managing multiple disjoint priority orders in priority queues

Abstract

In communication and computer systems, autonomous sources may assign priorities to their messages or jobs locally and independently. When a remote service (e.g., message transmission or RPC) is requested at a shared server, the server cannot use priority scheduling schemes effectively unless it can make a comparison between priorities defined by individual sources. The authors investigate the strategies under which the service received by requests of one source is not affected by the priority assignments at other sources. The first approach is a combination of processor-sharing and priority queue strategies. The second approach is to map locally defined priorities onto a global priority system. The performance of these approaches is examined in terms of the average response time of all requests, the average response time of the highest priority requests and a fairness measure. >