摘要
随着互联网的迅速发展 ,基于异构分布式网络的高性能计算平台——计算网格 ,成为了一种新型的计算模式 .介绍了基于 Globus的 How U网格的实现技术 ,通过资源请求代理进行网格作业提交 .为了有效利用网格资源 ,以“参数考察”任务为例 ,提出了一种自适应网格任务调度模型 ,该模型充分考虑了任务的实时容错需求 .最后比较分析了该模型与 Globus协作分配策略和 Gr
Advances in wide area network have led to the emergence of a new type of computing model, which is based on a high-performance distributed heterogeneous platform—computational grid. This paper has introduced the implementation of a Globus-based HowU grid system, and the client can submit its jobs via grid resource broker. In order to utilize grid resources effectively, this paper proposes a self-adaptable task scheduling model oriented parameter-study tasks. This model fully considers real-time, and fault-tolerant requirement of grid tasks. At last we have analyzed the performance of this model by comparing it with Globus co-allocated scheduling strategy and GrADS scheduling framework.
出处
《小型微型计算机系统》
CSCD
北大核心
2004年第11期1889-1893,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金资助 (60 12 5 2 0 8和 60 2 73 0 76)
关键词
计算网格
信息服务器
资源请求代理
自适应调度模型
computational grid
information server
grid resource broker
self-adaptable scheduling model