HMSST+:基于分布式内存数据库的HMSST算法优化

为了解决HMSST(HashMapSelectivityStrategyTree)算法在集中式环境下受限于有限内存的问题,提出了一种新的分布式SPARQL查询优化算法HMSST+。该算法基于Redis提出了一种分布式存储方案,通过平行扩展存储节点和分布式调度,使得海量RDF数据的查询得以在分布集群的内存中实现。采用LUBM1000所大学的测试数据集对查询策略进行了实验,结果表明提出的方法与HMSST算法相比具有更好的扩展能力,与现有的分布式查询方案相比也具有更好的查询效率。

HMSST+:HMSST Algorithm Optimization Based on Distributed Memory Database

To solve the bottleneck of HMSST(HashMapSelectivityStrategyTree) algorithm which is limited to the memory in a centralized environment,this paper proposed a novel distributed SPARQL optimized query algorithm named HMSST+.This algorithm presents a distributed storage solution based on the Redis(Remote Dictionary Ser-ver),and realizes the query of massive RDF data in the memory of distributed cluster by a parallel expansion of storage nodes and distributed scheduling .The method was tested on LUBM Benchmark and it worked well when the number of universities reaches 1000.The result shows that the method has better scalability than the HMSST algorithm and higher query efficiency than the existing query schemes.