有预算和截止期约束的网格作业调度算法

Scheduling Algorithm of Grid Jobs with Budget and Deadline Constraints

在商业网格计算环境中,作业有预算和截止期限制。如何向消费者提供有质量保障的服务,同时考虑服务提供者的利益,是一个关键问题。现有的作业调度算法只从消费者的角度出发对作业完成的时间和成本进行优化。同时从消费者和服务者的角度,利用作业的属性定义了作业的价值密度,在此基础上提出了高价值密度优先的网格作业调度算法HVDF。仿真结果表明,HVDF算法在实现价值率和按时完成作业数两个性能指标上优于现有算法。

In Grid computing environment,jobs have budget and deadline constraints.It is a key issue to provide services to Grid service consumers with QOS assurance and take the interests of providers into account at the same time.The existing scheduling algorithms of Grid jobs only try to optimize the costs and completion times of jobs from consumers＇ point of view.Both consumers and providers＇ interests were considered here.Value densities of Grid jobs were defined using jobs＇ parameters,and a scheduling algorithm of Grid jobs named HVDF（Highest Value Density First） was proposed.The results of simulation show that HVDF outperforms other algorithms in metrics of Hit Value Ratio and number of jobs finished before deadlines.