基于价值密度和截止期的网格作业调度算法

Algorithm of Grid Job Schedule Based on Value Density and Deadline

对商业网格中的作业调度问题进行研究,采用作业的到达时间、计算量、预算和截止期4个参数定义作业的优先级。在此基础上提出基于价值密度和相对截止期的网格作业调度算法,并对其进行仿真。仿真结果表明,该算法在实现价值率、按时完成作业数和加权作业按时完成率3个性能指标上优于现有算法,兼顾了消费者和服务者的利益。

This paper studies the job schedule problem in commercial grid computing. Four parameters of jobs, arrival time, computation, budget and deadline, are used to define a priority of jobs. A job schedule algorithm is proposed and simulated. Results of simulation show that this algorithm outperforms existing algorithms in metrics of Hit Value Ratio（HVR）, Weighted Job Fulfillment Ratio（WJFR）, and number of jobs finished before deadline. It achieves higher quality of service for consumers and higher income for providers than the contrastive algorithms.