【作者】 田原； 王营康； 肖达； 杨榆；

【Author】 Tian Yuan;Wang Yingkang;Xiao Da;Yang Yu;Information Security Center,Beijing University of Posts and Telecommunications;National Computer Network Emergency Response Coordination Center;

【机构】 北京邮电大学信息安全中心； 工业和信息化部信息中心；

【摘要】 随着如今用户对网络的依赖程度越来越高,用户对分布式数据存储的实用性与快捷性要求也变得越来越高。本文的研究主要集中在数据库存储技术与数据拆分算法的效率与速度问题,比较了基于Cassandra与Redis两种分布式数据库的存储效率,比较了Reed-Solomon(RS)码与Tornado码两种纠删码算法的冗余效率,最终设计仿真了一套适用于云存储的方案。实验仿真发现,同样是Key-Value分布式数据库,基于内存Redis的存储效率远大于基于硬盘空间的Cassandra,但是需要定时本地化存储,增加了存储开销。Reed-Solomon(RS)码的编解码效率低于Tornado码但是可靠性高于后者。通过比较分析,本文构造了不同要求下的冗余云存储系统,适用于不同的用户要求。更多还原

【Abstract】 According to the users are more and more depend on network nowadays,the utility and quick requirements of distributed data storage has become more and more high.This study focused on the efficiency and speed of database storage technology and data splitting algorithm,Comparison of storage efficiency between Cassandra and Redis,Comparison of the redundancy efficiency between Reed-Solomon(RS) code and Tornado code,conclude a plan for cloud system at last.The experimental results,Redis which based on memory is far greater than Cassandra which based on hard disk space on storage efficiency,but required more time to localized storage.Reed-Solomon(RS) code decoding efficiency is lower than that of the Tornado code but reliability is higher than the latter one.Through the comparative analysis,this study constructs the redundant cloud storagesystem under different circumstances suitable for different user requirements.更多还原