Cloud Computing Based Storage And Management On Spatial Vector Data

The increasing size of the spatial data and the complexity degree of the spatial analysis and processing technologies presents new challenges for the storage and management of massive spatial data. Infinitely scalable cloud storage capacity and computing power could meet the needs of mass spatial data storage, spatial big data processing, long-lasting online services. Based on so many advantages of cloud computing technology, the paper focused on how to use cloud computing technology to store and manage the vast amounts of spatial vector data. This paper studied on the index construction, data partitioning strategy, unique coding of the spatial vector data andits parallel import, query, update in a cloud environment. The following aspects of this paper are:(1) The background and related theories of the spatial vector data in the cloud storage. Firstly, this section analyzedthe application needsof spatial vector data storage and management from the massive spatial data storage, spatial big data processing, long-lasting online services, and proposed the research significance. Secondly,the current status of the domestic and foreign research were introduced. Moreover, basic theory of spatial vector data cloud storage from three aspects,including cloud computing, Hadoop distributed computing platform, storage and management of spatial vector data.(2) In the second section, spatial vector data storage applied to Hadoop cloud platform was proposed. According to the multi-scale characteristics of spatial vector data, multi-level grid index was designed based on the quadtree multi-scale segmentation technique.The partitioning way of spatial vector data was designed based on clustering properties of Hilbert spatial-filling curve.Combining spatial information multi-level grid coding and Hilbert coding, spatial vector data unique identificationwas designed conformed tothe RowKey storage rules. The storage rules of the spatial vector data, indexes, metadata, attribute data dictionary were designedaccording to HBase storage rule.(3) The third section achieved spatial vector data management based on Hadoop cloud platform. The framework and computing mode of the MapReduceparallel calculation model were analysed, and parallel importing process of spatial vector data was designed based on this MapReduce model.Then we designed the vector data parallel query general process based on MapReduce and ananlyzed the specific processes by examplesof parallel spatial selection query and parallel spatial KNN query. At last,we discussed the way to update the massive spatial vector data in cloud environments and designed two updating methodsfor local vector data and overall data separately.(4) Experiments were taken to verify the key technology. The prototype system to store and manage spatial vector data based on Hadoop are designed and realized the functions of spatial vector data import, query, update. Then we tested the efficiency of spatial vector data parallel import, the clustering effect of spatial vector data unique identification and the efficiency of spatial vector data parallel query.