Research And Implement Of Massive GIS Spatio-temporal Data Storage Method In Cloud Environment

With the development of mobile internet,sensor and big data technology,the amount of spatio-temporal data related to geographical information has increased explosively.More and more information data and applications are related to geographical space and time.The efficient storage and management of massive GIS spatio-temporal data has become more important.Spatial-temporal data itself has spatial attributes and temporal attributes.The traditional spatio-temporal data models are completely based on relational database,the service performance,data schema,and index performance are limited.The No SQL system in the cloud environment can handle the queries of large-scale spatio-temporal data,but the massive GIS spatio-temporal data has complex structures and are partly relevant,and the indexing technology provided by the No SQL system is not optimized for spatio-temporal data.This thesis proposes a massive spatial-temporal data storage proposal that is widely applicable to No SQL system,from spatiotemporal data modeling,read-write model architecture and spatio-temporal indexing aspects,considering the characteristics of spatiotemporal data.Aiming at the complex structure and relatedness of massive GIS spatio-temporal data,an ST-Open GIS spatio-temporal data model suitable for cloud environment storage and maintaining spatio-temporal data structure and relevance is proposed in spatio-temporal data modeling aspect.In this model,spatio-temporal data is abstracted to be an object with geospatial dimension,time dimension,and attribute dimension.The details of the spatial dimension,temporal dimension,and attribute dimension information,as well as temporal relationships and spatial topological relationships are introduced.The traditional GIS is facing the problems that poor read and write efficiency,lacks of scalability,flexible of data model and applicable index technology in relational databases.For the solution of mass GIS spatio-temporal data storage,a general model architecture for reading and writing massive GIS spatio-temporal data is summarized in three parts: client,No SQL interface layer,and No SQL data layer.Aiming at the weak space-time indexing ability of No SQL database,based on Map Reduce framework,a global distributed DBZ index of improved BZ tree is proposed.In this index method,spatio-temporal data is reduced by space-filling curve Zorder code,and then an efficient index is built by combining B+ tree index.The node information of index is stored in the No SQL database,which implements offline processing of data and indexes,on-line maintenance,and query strategies in order to realize high availability,high fault tolerance,high scalability,and high performance of index tables and data tables.Based on the above,this thesis implements a prototype system,verifies some abilities and tests performance.The results show that the prototype system is superior to the traditional relational storage architecture and can achieve the purpose of efficient abilities to storage and processing spatio-temporal queries.