摘要
针对现有技术中对小文件性能优化的效率低、运算不稳定等问题,研究出适用的解决方案。该方案基于OpenStack云平台构建软件定义对象存储系统,其中,分类合并处理模块采用分类合并方法和改进的GDS缓存替换算法,以实现缓存空间的合理分配与使用。动态预处理模块采用What-if假设方法并融合关联规则技术进行改进,提高对多种结构的小文件数据处理效率。试验表明,文中的存储系统中小文件的平均读写响应时间分别减少了32.3%和25.1%,适用性和可靠性高。
Aiming at the problems of low efficiency and instability in the performance optimization of small files in the prior art,suitable solutions have been developed.The solution is based on the OpenStack cloud platform to build a software-defined object storage system.The classification and merging processing module adopts the classification and merging method and improves the GDS cache replacement algorithm to realize the reasonable allocation and use of cache space.The dynamic preprocessing module adopts the What-if hypothesis method and integrates the association rule technology to improve the processing efficiency of small file data with multiple structures.The experimental results show that the average read response time of small files in the storage system of this study is reduced by 32.3%and 25.1%,respectively,which has high applicability and reliability.
作者
徐敏
胡聪
王萍
张翠翠
王鹏
XU Min;HU Cong;WANG Ping;ZHANG Cui-cui;WANG Peng(Information and Communication Branch of State Grid Anhui Electric Power Co.,Ltd.,Hefei 230061,China)
出处
《信息技术》
2021年第4期35-39,45,共6页
Information Technology
基金
国网安徽省电力有限公司科技项目(52120718004W)。
关键词
云计算技术
软件定义存储
小文件
分类合并
What-if假设
cloud computing technology
software-defined storage
small files
classification and merging
What-if hypothesis
