云存储中动态副本放置机制研究

发布时间：2018-03-11 19:16

  本文选题：云存储　切入点：副本　出处：《计算机工程与科学》2017年09期 　论文类型：期刊论文

【摘要】：数据副本管理是云计算系统管理的重要组成部分,在云计算系统的海量数据处理过程中,针对目前已知的数据存放与资源调度算法存在考虑副本动态性和可靠性的不足,提出了一种动态的副本放置机制。该机制基于区域结构,考虑数据处理时其副本的数量和放置位置,以及副本的产生对于内存和带宽等系统资源的开销:首先根据云存储中的副本信息,对被访问频率高且访问平均响应时间长的数据信息进行复制,并给出副本数量的计算方法;考虑缩小副本分布的节点选择范围,提出动态的副本放置算法——DRA,将一定范围内的节点根据提出的域的划分,进行放置筛选,以存放数据副本。实验结果表明,提出的动态放置机制不仅减少了低访问率副本对系统存储空间的浪费;同时也减少了高访问率副本所需跨节点的传输延迟,有效提高了云存储系统中的数据文件的访问效率、负载的均衡水平,以及云存储系统的可靠性和可用性。
[Abstract]:A copy of the data management is an important part of the management system of cloud computing, cloud computing in the massive data processing system, according to the known data storage and scheduling algorithm are considered insufficient copies of dynamic and reliability, proposed a dynamic replica placement mechanism. The mechanism is based on the regional structure, considering the data processing when the number of copies and location, and a copy of the memory bandwidth and the spending of system resources: Based on cloud storage in a copy of information, long data information of the accessed frequency is higher and the average response time to access the copy, copy number and calculation method are given; considering reducing the copy distribution of nodes the range of choices, propose a dynamic replica placement algorithm, DRA, will be within a certain range of nodes according to the division of the region, were placed on screening, to store a copy of the data . the experimental results show that the proposed dynamic placement mechanism not only reduces the rate of access to low copy system storage space; while reducing the transmission of high access rate copy required cross node delay, effectively improve the efficiency of accessing cloud storage system in the data file, the equilibrium level of negative load, and the reliability of cloud storage the system and availability.

【作者单位】： 东北大学信息科学与技术学院;
【基金】：国家自然科学基金(61374097)
【分类号】：TP333
，

本文编号：1599529