摘要
容错机制中基于任务数量的平均调度策略在处理跨度和服务质量方面存在不足,为此,提出一种基于优先级的网格调度算法,进而给出层次式集群系统的设计方案。在任务调度过程中引入任务剩余执行时间、任务价值密度、费用预算以及处理跨度的概念,以缩短任务处理跨度,提高服务质量。实验结果表明,与原机制调度策略和Max-Min算法相比,该算法在任务完成率、价值实现率和处理速率方面具有优势。利用该算法对原机制进行改进,能够有效提高系统的任务执行效率。
The average scheduling strategy based on task numbers in fault-tolerant has shortcoming such as processing span and Quality of Service（QoS）. Aiming at this problem,this paper proposes a grid scheduling algorithm based on priority,and then puts forward the design scheme of the hierarchical cluster system. The algorithm introduces the task in the process of task scheduling,task execution time remaining concept value density,cost budget and span,to make up for the task scheduling in the treatment of two span and QoS. Experimental results show that the algorithm has better effect in the completion rate,value realization rate,treatment rate compared with the original mechanism of scheduling strategy and Max-Min algorithm. The original mechanism is improved by using this algorithm. It can effectively improve the efficiency of system task execution,and the availability of system.
出处
《计算机工程》
CAS
CSCD
2014年第10期11-16,共6页
Computer Engineering
基金
广东省教育研究院教育研究课题基金资助项目(GDJY-2014-B-b243)
关键词
任务调度
价值密度
费用预算
处理跨度
高可用性
网格计算
task scheduling
value density
cost budget
processing span
high availability
grid computing
