A kind of the Cluster Computing Scheduling Algorithm Based on LAN
This paper discusses a parallel task-scheduling algorithm in network cluster computing system based on the model of multiprocessor parallel task scheduling. Compared with other methods, heuristic table scheduling can achieve better result with less cost. We designed this parallel task-scheduling algorithm for cluster system based on heuristic table scheduling algorithm. First, define the two properties of priority as: t-level(top-level) and b-level (bottom level). Then, Critical path is defined. Third, consider critical path node. Finally, we execute node-transferring scheduling. We apply a series of table scheduling algorithms, such as DLS (Dynamic level Scheduling,DLS) and MCP (Modified Critical Path) to compare performance. Our simulation experiment result shows this algorithm greatly improves scheduling performance.
cluster computing multiprocessor job scheduling Critical path Node-transferring scheduling algorithms
Qing Yang Yan Hu Ge Wang Shijue Zheng
Department of Computer Science, Central China Normal University, Wuhan, Wuhan, Hubei 430079, P.R.China
国际会议
杭州
英文
209-212
2006-10-12（万方平台首次上网日期，不代表论文的发表时间）