分布式目录失效的恢复算法研究

Research of failure recovery algorithm of distributed directory

提出了两种针对分布式目录失效情况时的恢复算法。基于元数据遍历的恢复算法采用遍历所有元数据的方式来近似地恢复目录,该算法能较完整地恢复目录。基于目录缓存的恢复算法采用目录缓存作为恢复的线索来近似地恢复目录,该算法能快速地恢复目录。两种算法在目录恢复上各有优缺点,通过实验验证了两种算法的功能,并对其性能做了比较。

Two failure recovery algorithms of the distributed directory are proposed. The recovery algorithm of the distributed directory based on Traversing Metadata traverses all metadata to recover the directory approximately. The recovery algorithm of the distributed directory based on Directory Cache takes the directory caches as a hint to recover the directory approximately and quickly. Both the algorithms have merits and drawbacks. Their functions are proved by experiments and their performances are compared.