节点文献
基于哈希规则的分布式文件系统的设计与实现
Design and Implementation of Distributed File System Based on the Hash Rule
【摘要】 随着全球数据量猛增,分布式文件系统逐渐成为存储技术的研究重点.文章设计与实现了基于哈希规则的分布式文件系统,它是一个可扩展的集群文件系统,并提供了一个具有两层结构哈希规则,规则为客户端提供了文件定位服务,其伪随机性也将文件均匀的分布到了不同的服务器中,起到了一定的负载均衡的作用;系统中采用桶来管理文件元数据与文件数据,桶的独立性与灵活性使得桶便于迁移,增加了系统的可扩展性;系统中同时采用了Lazy机制来达成系统性能与文件副本可用性之间的平衡.文中首先介绍了系统的整体架构,然后描述了系统中采用的关键技术与系统典型流程,最后,对系统的性能进行的测试证明了其高效性.
【Abstract】 With the increase of the amount of data around the whole world,distributed file system gradually becomes the key point of the storage technology research.The paper designed and implemented the distributed file system based on the hash rule.It is a extendible cluster file system,which provides a two-level structure hash-rule.The rule offers file location service for clients,and its pseudo-random feature makes file uniformly distributed on different servers.Thus the hash-rule plays a certain part in load balancing.Bucket is an entity that the file system uses to manage file data and metadata.The independence and the flexibility makes bucket easily to migrate,which increases the system scalability from another point of view.The system also uses the Lazy mechanism to achieve the balance between the performance and the duplication availability.We introduced the system architecture first.Then we described the key technique and the representative process of the system.The performance test of the file system in the last part proves its effectiveness.
- 【文献出处】 小型微型计算机系统 ,Journal of Chinese Computer Systems , 编辑部邮箱 ,2013年06期
- 【分类号】TP311.52
- 【被引频次】3
- 【下载频次】145