云存储上的I/O特征获取机制  

Efficient I/O characteristics collection method on cloud storage

作　　者：施杨斌[1] 吴杰[1] 梁瑾[1] 

机构地区：[1]复旦大学计算机科学技术学院,上海200433

出　　处：《计算机工程与设计》2011年第8期2870-2873,2888,共5页Computer Engineering and Design

摘　　要：为有效地获取云存储系统上的I/O特征,从云存储的体系结构出发,根据不同层面的结构特征提出一种层次化的I/O特征获取机制。该机制从全局云平台、集群和局部物理节点两个方面收集和分析云存储系统上的I/O特征数据,通过在云存储虚拟层嵌入监听模块来采集I/O特征,并通过统计分析模块对收集的数据做在线的分析。实验结果表明,该机制能够有效获取云存储系统的I/O特征,为云存储系统性能优化提供参考。To effectively collect the I/O characteristics of cloud storage system, a hierarchical I/O characteristics analysis mechanism based on the structure of different layers of cloud storage system architecture is proposed. In this mechanism, the UO characteristics are collected and analyzed from both global cloud platform and local physical nodes. The I/O characteristics are collected by embedded I/O monitor module in the virtual layer, and is analyzed by the on-line statistical analysis module. Experimental results show that this mecha- nism can effectively collect the I/O characteristics of the cloud storage system which provides reference for performance optimization of cloud storage system.

关 键 词：云存储 I/O特征 虚拟机 性能优化 云计算 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]