【摘要】 为解决数据网格中数据密集型作业的有效调度问题,提出了一种考虑等待队列中的作业潜在行为的基于访问代价的作业调度算法。分析了等待队列中的作业潜在行为的特点和作业访问代价的影响因素,给出了一个基于预测访问代价的作业调度算法。在OptorSim中的实验结果表明,该方法在作业平均处理时间和副本复制次数方面优于传统基于访问代价算法。更多还原

【Abstract】 To solve the scheduling issue of intensive data jobs in data grid,a job scheduling algorithm based on predictive access cost was proposed.Characteristics of potential behaviors of jobs in waiting queues were analyzed and the influencing factors to access cost were discussed.A job scheduling algorithm was given based on predictive access cost.The simulation result in OptorSim shows that it has a better performance in mean job time and number of replications than the traditional scheduling algorithm based on access cost.更多还原