一种支持并行处理的矢量数据存储与查询方法  

A vector data storage and query method supporting parallel compute

作　　者：陈洁[1] 褚龙现[2,3] 夏栋梁[2] 

机构地区：[1]铜川职业技术学院人文科学系,陕西铜川727031 [2]平顶山学院软件学院,河南平顶山467000 [3]武汉大学计算机学院,湖北武汉430072

出　　处：《电子设计工程》2017年第10期31-33,37,共4页Electronic Design Engineering

基　　金：河南省教育厅科学技术研究重点项目(12B520040);平顶山学院青年基金项目(PXYQNJJ2016013)

摘　　要：为了提高海量空间矢量数据的存储和拓扑关系查询效率,提出一种矢量数据的分布式存储与索引方法。设计了基于HBase的空间数据存储模型和索引构建方案,采用Spark计算框架实现了网格空间索引的并行构建算法,利用索引完成了空间拓扑关系的分布式查询。最后在Hadoop集群上统计了相同数据集的拓扑包含查询时间,结果表明提出的并行存储与查询方法可行性好,比直接查询HBase算法快4~5倍。To improve the storage efficiency of massive spatial vector data and query efficiency of topotaxy relation, a distributed storage and index method of vector data is proposed. A spatial data storage modal and index build method is designed based on HBase. The parallel build algorithm of grid spatial index is realized by using Spark compute framework, and the distributed query of spatial topotaxy relation is then achieved by using the index. The query time consumption of topologically containing based on same data sets on Hadoop cluster is counted, statistic data shows that the proposed method on parallel storage and query is feasible and 4-5 times faster than directly query method of HBase.

关 键 词：空间关系 并行 SPARK HBASE 区域查询 

分 类 号：TN91[电子电信—通信与信息系统]