基于大规模分布式副本定位的分级索引压缩机制

Novel Hierarchical Index Compression Mechanism Based on Large Scale Distributed Replica Location

针对超级节点索引方式下的大规模分布式系统,提出一种用于副本定位的资源索引分级压缩机制。该机制把超级节点所辖分级网络中上层节点的有序子节点集映射到一个位串向量,进而通过自下而上的索引发布和索引在上级节点的汇聚实现冗余副本记录数的压缩,副本定位则通过逆向的位串查询实现。实验表明,该机制可达到较高的记录压缩比,并在一定程度上提高副本定位效率。

Aiming at super-node indexed large scale distributed system（LSDS）,a kind of novel hierarchical index compression mechanism is put forward for replica location.It is implemented by mapping all sub nodes of higher-up node to corresponding bit string at first,then compressing record number of redundant replica index by bottom-up index publish and index aggregating on higher-up node.Accordingly,replicas in system can be located by query on bit string conversely.The experimental results indicate that this mechanism achieves high record compression ratio and has some active impacts on efficiency of replica location in LSDS.