网络编码在分布式存储系统中运用的研究

发布时间：2018-04-01 07:55

本文选题：网络编码　切入点：分布式存储　出处：《西南交通大学》2014年硕士论文

【摘要】：网络编码由R.Alswede于“Network Information Flow”一文中首次提出,它彻底改变了传统路由的存储转发模式,允许中间节点对收到的数据包进行信息重组,从而达到多播通信网络的容量上界。Acedanski等人在“How good is random linear coding based distributed networked storage“一文中更是将网络编码的思想引入到存储技术中,提出了一种随机线性编码方案。网络编码自提出以来就受到了各界学者的密切关注。近年来,随着互联网的飞速发展,网络信息量呈指数级的增长,信息技术产业已将其技术重心转移至对存储技术的研究。传统的客户机/服务器存储模型已无法满足人们日益增长的存储需求。分布式存储主要利用网络中闲置的存储空间,通过网络技术将网络中的闲置PC机组合起来搭建成信息存储资源池。因其良好的可扩展性和低廉的价格得到了广泛的运用。有人提出将网络编码技术运用于分布式存储系统中。研究表明,将网络编码应用于分布式存储系统不仅能够节约带宽消耗,还能很好的实现负载均衡,增强系统的健壮性,对分布式存储系统的发展和改进意义重大。本文首先对Acedanski等人提出的基于随机线性编码的存储方案进行了研究,对比了复制、纠删码等常见存储方案,得出随机线性编码存储方案能在消耗更少额外存储带宽的情况下以较高的概率完成源文件重构的结论。接着,本文针对网络编码方案对分布式存储系统下基于再生码的节点修复技术进行了研究。该修复技术最大的优点在于无需下载整个源文件就可实现失效节点的修复,能大大减少节点修复时的带宽开销。相关文献证明了再生码方案存在两个极值点即：最小存储带宽再生点(Minimum storage regeneration points, MSR)和最小节点修复带宽再生点(Minimum bandwidth regeneration points, MBR)。本文对其对应的两种编码修复技术：最小存储再生码(Minimum storage regeneration code, MSRC)和最小带宽再生码(Minimum bandwidth regeneration code, MBRC)进行了深入的研究。通过本文的数值计算结果表明基于MBRC的编码修复技术能在节点修复时实现带宽消耗的最小值。文章最后对基于网络编码的MBRC方案进行了仿真实现,基于VC++6.0平台对MBRC方案的数据分布、数据重构、失效节点修复几个过程进行了模拟仿真。利用仿真结果验证了MBRC方案的有效性。
[Abstract]:Network coding is first proposed by R.Alswede in "Network Information Flow". It completely changes the traditional routing mode and allows intermediate nodes to reorganize data packets. Acedanski et al introduced the idea of network coding into storage technology in "How good is random linear coding based distributed networked storage". In this paper, a random linear coding scheme is proposed. Network coding has been paid close attention by scholars since it was put forward. In recent years, with the rapid development of the Internet, the amount of network information is increasing exponentially. The information technology industry has shifted its technology focus to the research of storage technology. The traditional client / server storage model can no longer meet the increasing demand for storage. Distributed storage mainly makes use of the idle storage space in the network. The idle PC in the network is combined to form a pool of information storage resources through network technology. Because of its good expansibility and low price, it has been widely used. Some people have proposed to apply network coding technology to distribution. Research shows that, The application of network coding in distributed storage system can not only save bandwidth consumption, but also achieve load balance and enhance the robustness of the system. It is of great significance to the development and improvement of distributed storage system. In this paper, the storage schemes based on random linear coding proposed by Acedanski et al are studied, and the common storage schemes such as replication, erasure code and erasure code are compared. It is concluded that the stochastic linear coding storage scheme can complete the reconstruction of source files with a higher probability of consuming less additional storage bandwidth. In this paper, the node repair technology based on regenerative code in distributed storage system is studied according to the network coding scheme. The biggest advantage of this repair technique is that the failure node can be repaired without downloading the whole source file. It is proved that there are two extremum points in the scheme: minimum storage regeneration points (MSRs) and minimum bandwidth regeneration points (MSRs). Two corresponding coding repair techniques, minimum memory storage regeneration code (MSRC) and minimum bandwidth regenerative code minimum bandwidth regeneration (MBRC), are studied in detail. The numerical results of this paper show that the MBRC based code repair technique is based on MBRC. The minimum bandwidth consumption can be realized when nodes are repaired. Finally, the MBRC scheme based on network coding is simulated. The data distribution, data reconfiguration and failure node repair of MBRC are simulated based on VC 6.0 platform, and the validity of MBRC scheme is verified by simulation results.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP333

【参考文献】