海量存储系统中的元数据分级索引算法*

海量存储系统中,高效的元数据索引是减少查找元数据所需时间与空间开销的重要手段。针对现有元数据管理方法存在查找元数据所需时间与空间开销大和性能波动大等问题,设计了元数据分级索引算法。依据元数据的生命周期,将元数据分为活跃和非活跃两级;使用Bloom Fliter对均衡的活跃元数据分区生成摘要串,并使用B-树建立活跃元数据分区的索引;使用类似的方法对非活跃元数据分区,并为每个分区选择各自的哈希函数。从查找元数据所需时间与空间开销、适应能力两方面对元数据分级索引算法进行了分析,并与现有元数据管理算法进行了比较。最后实现了元数据分级索引算法的原型系统,使用真实数据集进行了测试与分析,结果表明,元数据分级索引算法能减少查找元数据所需的时间与空间开销,并具有很强的适应能力。

Hierarchical metadata indexing algorithm of mass storage system

Efficient metadata index is important to reduce the time and space overhead of metadata management in mass storage system.Current metadata management algorithms need large and volatile time and space overhead,this paper presented a hierarchical metadata indexing algorithm.Based on the life cycle of metadata,classified all metadata into active metadata and non-active metadata.The active metadata was partitioned,used the Bloom Filter to generate binary string of each partition,and used the B-tree to establish index of each active metadata partition.Used the similar approach to partition the non-active metadata and the suitable hash function was dynamically selected for each non-active metadata partition.The metadata query time and the algorithm’s capability were analyzed and compared it with state of the art metadata management algorithms.Finally,implemented the prototype of hierarchical metadata indexing algorithm.The experimental results show that the proposed algorithm can reduce the time and space overhead of metadata query and has high adaptability.