Storage And Parallel Query Technology Research In Distributed Environments Massive Spatial Data

With the wide application of GIS in every industry, plus people s precisionrequirement toward spatial data increases constantly, the spatial data of GIS whichneeds management becomes more and more,the data volume also is becoming biggerand bigger. Recently, The development of maturity of distributed storage and parallelcomputing provide a brand new direction for storage and processing of massivespatial data.However,in the distributive environment the current method for storageand processing of massive spatial data are mostly based on the traditional relationaldatabase. Due to its particular relation mode, the fast access and treatment ability formassive spatial data is limited, so it is not an ideal database for two or threedimension spatial data storage. So how to utilize the non-relational database to realizethe efficient storage and the fast processing under distributed environment possess animportant research significance.Firstly, according to comparatively analysing the characteristics of the currentcategories of typical non-relational database, we put forword advantages of non-relational database MongoDB in processing the massive data. After summarize thecharacteristics of spatial data as stated above, we design the style of storage ofmassive spatial data in MongoDB, and do some comparison experiments in dataimport and data query with relation database SqlServer for verfing feasibility ofMongoDB in storage capacity. According to style of storage above, we combineMapReduce which belongs to the parallel processing framework Hadoop to realizeparallel processing of massive spatial data.Finally, we design and realize the databasemanager HMGIS which is based on Hadoop and MongoDB, and set up the distributedenvironment of HMGIS in couples of Server machines with Linux for verifingfeasibility and efficiency of HMGIS in processing by massive spatial data import anda variety of spatial queries.The main research results of thesis include:(1) It’s feasible that we use MongoDB to store the unstructed massive spatialdata；(2) the database manager HMGIS which is based on Hadoop’s MapReduce andMongoDB database has higher accessed and queried efficiency than traditionalrelational database in processing spatial data;(3) implementation of HMGIS can provide software support for the massive spatial data analysis and complex Geographic computation.