可扩展的分布式矢量空间数据库集群原型系统研究

空间数据获取手段呈现多样化,其数据集每天以PB级的速度在增长,如何高性能地存储、高效处理海量空间数据成为重点问题。基于空间数据库集群系统,提出了Geohash的矢量空间数据分片存储方法,通过该分片方法实现了空间数据的并行导入、分布式矢量空间数据查询。通过实验分析了分布式矢量空间数据库在真实数据集以及虚拟数据集下的读写性能。实验表明:考虑空间分布特征的分布式空间数据库集群在空间查询性能和并发访问性能方面具有更好的扩展性。

Research on Prototype System of Scalable Distributed Cluster Vector Spatial Database

Because of the diversification of spatial data acquisition technologies,the volume of spatial data is growing exponentially.How to store and process these spatial data becomes an important problem for geospatial data applications.Based on a shared-nothing spatial database cluster,a Geohash based data partition method is put forward.Parallel importing and distributed query of vector data are implemented based on this data partition.The read/write performance of distributed cluster spatial database is analyzed.Experimental results on a real dataset and synthesized dataset showed that distributed cluster spatial database for large data set has a better performance in query and concurrent access.