Research On Massive Spatial Data Distributed Storage And Parallel Processing Technology

The rapid progress of the spatial data makes the spatial analysis and process technologies more complicated, meanwhile making it more difficult to efficiently manage and process the massive spatial data. Thus, there exists urgent need to new technologies and methods to manage and process massive spatial data. Fortunately, the distributed storage and parallel processing technologies provide us new way to solve the abovementioned problem.To solve the several shortcomings of the contemprary spatial data storage and process methods, based on summarizing the real application requirements, this thesis mainly studies the massive spatial data management technology based on HBase and the parallel processing technology. Main works of this thesis include:(1)Spatial data storage and parallel processing technologies are studied, which provide the theoretical basis for the whole thesis. Moreover, the shortcoming of traditional methods are analyzed and presented.(2)HBase which is based on distributed file system is studied. It forms the basis of our distributed storage and management method. A parallel index for massive spatial data is designed, meanwhile HBase based spatial data management method is proposed. Experimental results show that compared to traditional methods, HBased based method has more superiority in distributed storage performance.(3)Map ReduceGIS which is a parallel processing engine for massive spatial data is designed and implemented based on Map Reduce parallel programming framework. Because Map Reduce doesn't support spatial data join operation, a parallel spatial join method is designed.(4)Parallel processing experiments are carried out based on Map ReduceGIS. Experiments on the add, read, nearest neighbor query operations show that Map ReduceGIS has better performance in massive spatial data parallel processing than the traditional PostGIS system.