中文摘要：

针对气象水文应用中,大量常规观探测报文批量访问出现的低效问题,研究文件存储特性,定量分析了目录级数和文件数量对访问性能的影响,发现文件数相对于文件大小,对于系统的访问效率影响更大,当单个目录下文件数目过大时,文件存取延时较大,严重影响用户体验与服务性能。根据NTFS下的实验数据,设计了一种高效的目录组织方法,优化用户态文件存储管理算法。实验表明,优化后的文件目录结构和组织形式,能极大地提高批量文件的读取效率,降低20%—73%的访问延时,改善网络环境下的大规模文件接收处理效率。

英文摘要：

In meteorological and hydrological applications,batch file would lead to inefficient access.To solve this problem,we studied the file storage feature,and the access performance influenced by depth of directories and number of files through quantitative analysis.We found that with respect to the file size,file number had a greater impact on access performance.A large number of documents would result in a great time delay,which seriously affected the user experience and service performance.This article designed an efficient directory organization methods due to the data of experiment in NTFS,which optimize the user file storage algorithm.Tests showed that it significantly improved the efficiency of file access with 20% to 73% decrease of the latency, and had a high practical application value in massive files on the network environment.