分布式文件系统数据读写流程分析与优化-计算机系统结构专业论文.docx

华中科技大学硕士学位论文 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract In the big data era, storage systems play more and more important role in many practical applications. The efficiency of storage system directly determines the performance of the application running on the system. Now, most file systems fit the increasing need of performance by enhancing their scales, which also leads to some problems, such as more and more costs and the difficulty of safeguard. The object-based file systems delegate responsibility to object server, but they ignore one fact that all nodes in a system are an organic whole. The key factor of a successful storage system lies in making full use of every nodes and interconnection network in the system. Storage systems play more and more important role in many practical applications. The efficiency of storage system directly determines the performance of the application running by the system. The key factor of a successful storage system lies in making full use of every nodes and interconnection network in the system, which is also the key factor of improving the systems scalability and availability. All successful systems can make full use of all their components. In this thesis, we focus on the process of data I/O and optimize the key factor. The design and implementation base on Cappella which is an object-based distributed file system developed in our lab. In the process of writing data, we develop a dynamic scheme of data layout which based on real time workloads of the data sever. Every server has a weight standing for its business degree which is collected every 3 seconds. Before storing data, it is the weight that determines which one is selected. The dynamic data layout scheme successfully settles the problem of static one which is adopted in Cappella. In the process of reading data, we make an analysis to the linux kernels prefetching strategy and propose a method fitting in distributed environment to overcoming linux kernels drawback. Linux ke