基于网络编码的二层分布式存储及数据修复算法

发布时间：2018-04-16 22:15

本文选题：网络编码 + 分布式存储系统　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：随着互联网的发展与宽带的普及,数据量日益爆炸式的增长,分布式的数据存储采用可扩展的系统结构,使用多台存储服务器进行数据的存储降低了每台存储服务器的存储负荷,存储信息通过定位服务器进行定位,分布式存储方式不仅提高了系统的可靠性且其可拓展的系统结构满足海量数据的存储要求。分布式数据存储是数据收集型无线网络的关键技术之一,传统的基于数据备份的数据存储方法因存储开销大、无线节点可靠性差等原因不适合于分布式的无线网络,而将网络编码引入到分布式存储中,经证明在冗余度相同的情况下,相比于数据备份,基于网络编码的分布式存储可以取得更大的数据可靠性。虽然基于网络编码的分布式存储,可以以最小的存储开销获得最大的数据可靠性,但网络编码的引入,也引起了因编码数据失效而造成的数据修复问题。在基于网络编码的分布式存储中,为保持数据的可靠性,需对失效的编码数据进行数据修复。而传统的数据修复算法将产生巨大的通信开销,造成网络资源的浪费。为减少数据修复通信开销,本文提出了一种基于二层数据结构的分布式数据修复模型和方法。此方法以增加一部分存储开销为代价,降低数据修复所产生的修复开销。相比于MSR(Minimum Storage Regenerating Codes),基础的干扰联合和组干扰联合方法,本文所提出的修复方法具有更少的通信开销,经证明可证明所提议的方法可将数据修复开销降低到传统数据修复算法的o(1/(?))倍,适合于资源受限的无线网络分布式存储。
[Abstract]:With the development of the Internet and the popularization of broadband, the amount of data is increasing explosively. The distributed data storage adopts extensible system structure.The use of multiple storage servers for data storage reduces the storage load of each storage server, and the storage information is located through the location server.Distributed storage not only improves the reliability of the system, but also the extensible system structure meets the storage requirements of massive data.Distributed data storage is one of the key technologies in data collection wireless networks. Traditional data storage methods based on data backup are not suitable for distributed wireless networks due to high storage overhead and poor reliability of wireless nodes.When network coding is introduced into distributed storage, it is proved that the distributed storage based on network coding can achieve greater data reliability than data backup when redundancy is the same.Although distributed storage based on network coding can obtain maximum data reliability with minimum storage overhead, the introduction of network coding also causes data repair problems caused by the failure of encoded data.In distributed storage based on network coding, the invalid coded data should be repaired in order to maintain the reliability of the data.The traditional data repair algorithm will generate huge communication overhead, resulting in a waste of network resources.In order to reduce the communication overhead of data repair, this paper presents a distributed data repair model and method based on two-layer data structure.This method reduces the cost of data repair at the cost of increasing part of storage cost.Compared with MSR(Minimum Storage Regenerating Codesh, the proposed method has less communication overhead than MSR(Minimum Storage Regenerating Codesh, and it is proved that the proposed method can reduce the cost of data repair to 1 / 1 of the traditional data repair algorithm.It is suitable for distributed storage in wireless network with limited resources.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【相似文献】