学海网

数据网格中数据复制关键技术的研究电子书
摘要
数据网格的目标是使地理上广泛分布的团体能够共享数据资源，以完成对大规模海量数据集的复杂分析、处理任务。而数据复制技术是数据网格系统中广泛采用的改善网格系统性能的关键技术之一。本文对国内外数据网格数据复制技术现状进行了深入研究，并在此基础上对数据复制过程中的副本创建、副本选择等关键性问题各提出了一种新的解决策略，同时给出了其相应的算法。首先，研究了当前数据网格中基于用户驱动的副本创建策略以及基于经济模型的副本创建策略，分析了他们各自适用的环境。同时，根据当前数据网格网络带宽状况提出了一种网格集群分类结构，并在此基础上提出了基于网络带宽差异的副本创建策略HRS。实验结果显示，在网络可用带宽较低的情况下，HRS 策略和其他策略相比在作业执行效率上有显著提高。其次，针对数据网格资源配置和真实的市场资源配置的相似性，将价值机制以及博弈论的相关知识引入到副本选择策略当中，创建了副本选择博弈模型GMDS。最后，在 GMDS 模型的基础上提出了基于纳什均衡的数据网格副本选择策略DSSNE。实验结果显示，DSSNE 策略在执行作业较多、用户对副本请求较频繁的情况下和传统的副本选择策略相比在作业执行效率上有显著提高。
关键词 数据网格；复制策略；副本创建；副本选择；纳什均衡

目录
摘要 ······················································································································· I
Abstract··············································································································· III
第 1 章绪论········································································································· 1
1.1 研究背景··································································································· 1
1.2 国内外研究现状······················································································· 3
1.3 本文研究的内容······················································································· 5
1.4 研究意义··································································································· 6
1.5 本文组织结构··························································································· 7
第 2 章数据网格数据复制技术相关知识·························································· 9
2.1 数据网格体系结构··················································································· 9
2.1.1 五层沙漏体系结构············································································ 9
2.1.2 开放网格服务体系结构OGSA························································11
2.1.3 WEB 服务资源框架WSRF··········