摘要
分布式集群系统是云计算环境的基础,针对当前云计算环境高性能计算的需求,更好地发挥系统的运行效率,本文在分析了已有网格计算环境下分布式任务分配方法的基础上,从软件角度采用动态负载均衡思想,依据网络节点计算力模型,提出了海量数据集群环境计算的四叉树任务分配策略。验证表明,本任务分配策略能有效地提高整体计算速度。
Distributed cluster system is the foundation of cloud computing,aiming to the requirements of high performance computing for cloud computing environment,improving the efficiency of system implementation,this paper analyzed distributed task allocation methods on the basis of the existing grid computing environment,and adopting task allocation strategy of dynamic load balance from the field of software,and the ideas of parallel computing and estimation about the power of computing on network node,proposed a policy of task allocation within the condition of cluster to big data.It was proved that the conclusion of the strategy on task allocation can enhance the overall computing performance effectively.
出处
《电信科学》
北大核心
2010年第10期30-34,共5页
Telecommunications Science
基金
国家"863"计划基金资助项目(No.2007AA12Z182
No.2009AA12Z222)
浙江省重点攻关基金资助项目(No.2009C33011)
浙江省自然科学基金资助项目(No.Y5090130)
教育部博士点基金资助项目(No.200803350017)
关键词
分布式集群
高性能计算
计算力模型
四叉树结构
distributed cluster
high performance computing
power of computing model
quad-tree structure
