期刊名称：International Journal of Distributed Sensor Networks
印刷版ISSN：1550-1329
电子版ISSN：1550-1477
出版年度：2009
卷号：5
DOI：10.1080/15501320802505960
出版社：Hindawi Publishing Corporation
摘要：Task arriving time of a heterogeneous distributed system is stochastic. In order to make the heterogeneous distributed system satisfy user Qos (Quality of Service) requirement and transact tasks as much as possible and keep load balance, the Qos Min-Min scheduling strategy is described, which is composed of urgent Qos Min-Min scheduling strategy and high priority Qos Min-Min scheduling strategy and periodic Qos Min-Min scheduling strategy. In this paper, tasks are classified as urgent, high priority, and common task. The properties of urgent task are when it arrives at the system, it should be executed immediately and cannot be discarded by the system. The properties of high priority task are when it arrives at the system, it should be assigned and executed by the system as soon as possible without affecting elimination of the assigned task. Tasks other than the above two are defined as common tasks. When the urgent task arrives at the system, it will be assigned immediately to the system by urgent Qos Min-Min scheduling strategy, so urgent Qos Min-Min can guarantee that urgent task can be executed by the system. When high priority task occurs, it will be assigned immediately to the system by high priority Qos Min-Min scheduling strategy, so high priority Qos Min-Mint can reduce the ratio of the task discarded. Periodic Qos Min-Min scheduling strategy inserts common tasks into an appropriate position of the exact queue in the system, so periodic Qos Min-Min can also reduce the ratio of task discarded. Compared with Min-Min schedule strategy, Qos Min-Min can reasonably reduce the ratio of task discarded and improves utilizable CPU ratio. Implementing results show that the performance of Qos Min-Min is better than Min-Min.
