多因素感知的网格资源度量方法与任务调度

Grid resource metrics of multi-factor perception and task scheduling

资源管理和任务调度是网格计算的核心问题,目前这方面的研究已取得丰硕的成果,但是,在网格资源评估方面仍缺乏行之有效的途径.文章论证了任务调度中候选资源选择的重要性,提出了资源定量评估的线性组合方法,在此基础上提出了一种结合节点计算能力、负载量、带宽、通信距离以及费用等多种因素的网格资源度量方法,建立了相应的数学模型,并设计了新的任务调度算法,实验表明该算法有效改善了网格的性能.

Resource management and task scheduling are the core issue of grid computing. Currently, research in this area has achieved fruitful results. However, the grid resource assessment lacks effective way. This paper demonstrated the importance of the choice of candidate resources in task scheduling. Presenting a linear combi- nation method of quantitative assessment of resources, the paper proposed a grid resource metrics combination of various factors such as computing capacity of node, load, bandwidth, communication distance and cost. Establis- hing a corresponding mathematical model, we then designed a new task scheduling algorithm. Experimental re- sults showed that the algorithm effectively improved the performance of grid.