Cloud Computing Environments Of Large GML Spatial Data Parallel Access Key Technology Research

With the maturation of earth observation, mobile GIS, Internet, Internet of Things and other technologies, and expanding field of use GML, GML spatial data is a growth spurt, the amount of data being jumped from GB to PB and the EB-grade level, large GML data era is coming. The calculation of the traditional spatial database and I/O capacity can not meet the high performance required large GML data processing requirements. In recent years, popular cloud computing technology with ultra-large-scale, high scalability, high reliability and versatility features; while NoSQL big data in the background of vibrant, it is at the same time support for structured and semi-structured data stored in a kinds of non-relational distributed database. Therefore, cloud computing and NoSQL technology for large-scale parallel semistructured spatial data access issues GM gives a new solution.In this paper, the use of open source cloud computing platform Hadoop and HBase distributed database technology to study the key big GML data in a distributed computing environment, concurrent access. With Hadoop HBase major platforms and software for storage and query large GML spatial data, do the following research.(1) Analysis of the existing spatial data partitioning algorithm, combining traditional GML GML data storage management ideas and have the characteristics of both the integrity of the geographic features geometric and topological relations, and research under the cloud computing platform for large GML spatial data dynamic partitioning algorithm strategy.(2) with the existing cloud computing technology and NoSQL technology, designed for a storage model is stored in HBase GML data; analysis of distributed file system architecture and a copy of Hadoop data placement strategy to improve the default HDFS data copies placement, and distributed file system so that it can be expanded to ensure the integrity of the geometry of geographic features.(3) In-depth study of traditional spatial data indexing mechanism, integrated parallel spatial indexing algorithms and GML data partitioning algorithm, based on quad-tree and Rtree indexing mechanism designed a GML data for the two-stage parallel spatial index structure.(4) Analysis of the characteristics possessed by GML data and MapReduce parallel computing model; and HBase database query optimization techniques combined with traditional spatial database; then when considering partitioning strategy adopted GML data distributed storage for large GML cloud platform under study Spatial data parallel query algorithms and strategies.Finally, analyze the performance and efficiency through laboratory tests results GML data partitioning algorithm, storage model, parallel indexing mechanisms and query algorithms designed to have a good performance.