Research On Key Technologies Of Vetcor Data Storage And Query Processing Towards The Distributed Spatial Database

With the development of Space-Air-Ground integration,the technologies of geographic data acquisition are increasing,and the scale of vector data accumulated by all walks of life is exploding.The traditional service model which is based on a single-node spatial database has been difficult to meet the need for storage and query processing of large-scale vector data.Therefore,a high-performance distributed vector spatial database supporting a variety of spatial databases needs to be constructed to meet the requirements for distributed storage and efficient parallel query processing of large vector data in different application scenarios.Focusing on this goal,this thesis conducts in-depth research on several related key technologies including the vector data storage organization model,the vector data partitioning strategy and the distributed geographic structured query language,and then verifies the feasibility of these key technologies through the prototype system.Specific contents are as follows:(1)The design of distributed storage and organization model for vector data based on object-relation spatial database.In order to meet the storage requirements of vector data in a distributed spatial database,based on the research of the existing vector data models,this thesis combines data distribution characteristics under distributed environment and the idea of logical organization with layering and partitioning,and fully considers the advantages of object-relation spatial database in storage and query processing of geometric elements,and finally designs a distributed storage and organization model for vector data based on object-relation spatial database.(2)The research on vector data partitioning strategy based on Hilbert ordering code and jump consistent hash.In order to improve the efficiency of distributed storage and parallel query processing for large vector datasets,through analyzing the shortcomings of existing vector data partitioning methods,this thesis proposes a new vector data partitioning strategy based on Hilbert ordering code and jump consistent hash.The new strategy can build vector data blocks on demand,and allocate the data amount according to the performance of the service nodes.Furthermore,it also takes into account the data migration problems occurring in the expansion of system nodes.(3)The design of distributed geographic structured query language--DGSQL3.By comparing and analyzing the differences in query language between different object-relational spatial databases,as well as considering the characteristics of distributed queries,a distributed geographic structured query language--DGSQL3 is designed and implemented.It provides a global unified query interface for constructing a distributed vector spatial database management system that supports multiple spatial databases(PostGIS,MySQL Spatial,SQL Server Spatial).(4)The implementation of prototype system and the validity test of key technologies.Based on the above key technologies,a distributed vector spatial database prototype system is designed and constructed.The prototype system is applied to evaluate the effectiveness and related performance of those key technologies.The results show that the distributed vector spatial database constructed on the key technologies proposed by this thesis can achieve the distributed storage and efficient parallel query processing of large vector datasets,and support unified query of heterogeneous spatial databases.Meanwhile,the prototype system can remain excellent load balance and scalability under multi-core heterogeneous environment.