分布式存储平台的路由算法和本地资源管理

【作者】 李克南；

【导师】 卢显良；

【作者基本信息】 电子科技大学 ， 计算机软件与理论， 2004， 硕士

【摘要】 本课题来源于四川省青年软件创新工程项目:基于广域网的分布式容错存储平台。研究和开发一种网络容错存储平台软件,主要用于数据的高可靠存储、远程备份和远程灾难恢复。本文的核心内容是这个存储平台的路由算法和本地资源的管理。在本论文中,首先叙述了有关网络协议和服务器模型的基础知识,这些基础知识是研究P2P分布式存储模型的前提。接下来简单的总结了P2P分布式存储模型的历史和发展。接下来第五章描述了基于广域网的分布式容错存储平台的系统结构,从总体上明确了本论文的核心内容在整个系统中的功能和作用。第六章描述的是文件名到节点号的映射安全散列算法,作为一个Hash算法,它不仅提高了效率,而且用来在分布式系统中定位节点还达到了一定的负载平衡的作用,这一部分是后面描述路由算法的基础。第七章分析介绍了传统的路由算法,总结了它们的优点和不足之处,之后在第八章中提出了我们的一种面向层次结构的路由算法,这种面向层次结构的路由算法定位节点快,健壮性高,安全性好。第九章描述了本地资源管理,重点在于文件存储指针表的设计和管理,特别是保持内存和磁盘信息的同步机制,包括定时同步和补救同步。最后通过对基于广域网的分布式容错存储平台功能和性能的测试,表明它达到了设计的目标并具有较高的性能。更多还原

【Abstract】 Research & Development of distributed fault tolerant storage flat based on WAN is a project comes from “Youth Software Creative Engineering” from Technological Office of SiChuan. It mainly researches a network fault tolerant storage flat software, it can be used in high reliable storage of data、long-distance copy and calamity remedy. The most important parts of this paper are the Route algorithms and local resources management of this storage flat.Firstly, we recount some basal knowledge about network protocol and server model. Because these are the preconditions of researching P2P. In the end we draw a simple conclusion of the history and development of the P2P distributed storage model.Then, we discuss the system structure of distributed fault tolerant storage flat based on WAN, so we can understand the function and effect of this paper’s nuclear content in the whole system from a holistic view. In chap.6 we depict a secure hashing algorithm about mapping from file name to node number. As a hashing algorithm, it not only enhances the efficiency but also plays a load balance role in positioning node in the distributed system. The latter part is the base of the Route algorithm. In chap.7 we introduce a traditional Route algorithm and point out it’s advantages and disadvantages. In the following chap. a hiberarchy-oriented Route algorithm is presented, the prominent merit of this new algorithm is it’s fast positioning node、high reliability and well security. In chap 10. we describe the local resources management , the emphases are the design and management of the file storage pointer table, especially keeping the information in memory and disk synchronous which include timer synchronization and remedy synchronization。Finally, we proved that the initial goal is reached and a good performance is achieved by taking the system in some functionality tests.更多还原