摘要： 嵌入式服务器相对于传统服务器具有一定的特殊性和专用性,传统的文件系统难以满足特定嵌入式平台高效存储的需求。针对某特定嵌入式平台下大容量流式数据存储效率低下的问题,提出一种面向大容量流式数据的嵌入式文件系统SEFS,其中创建文件时借助三层树状统计结构为新文件快速预分配存储空间,采用基于磁盘饱和度的均衡存储策略。实验结果表明,SEFS创建单个文件用时59 ms,与Ext2和MSFS文件系统相比,读文件速率分别提升了8.63%和6.45%,写文件速率分别提升了6.48%和2.94%,并且具有较好的存储均衡性和稳定性。
关键词:
大容量,
流式数据,
嵌入式文件系统,
预分配,
磁盘饱和度
Abstract: Embedded servers have some specificities compared with traditional servers.Traditional file systems can no longer meet the demands of efficient storage for some specific embedded systems.This paper proposes a Streaming Embedded File System (SEFS),which is oriented to large capacity of streaming data,for a particular embedded system to solve the problem of low efficiency storage.In SEFS,storage space is rapidly pre-allocated in virtue of a three-layers tree structure of statistics while creating new files.In addition,it applies disk saturability-based strategies to guarantee storage balance of SEFS.Testing results show that this system creating a new file by SEFS costs 59 ms,the reading rate of SEFS is faster than Ext2 and MSFS by 8.63% and 6.45% respectively,and the writing rate of SEFS is faster than Ext2 and MSFS by 6.48% and 2.94%.Testing results also demonstrates SEFS has better storage balance and stability.
Key words:
large capacity,
streaming data,
embedded file system,
pre-allocation,
disk saturability
中图分类号: