云计算环境下GML空间数据存储索引机制研究

发布时间：2018-01-03 06:29

本文关键词：云计算环境下GML空间数据存储索引机制研究　出处：《江西理工大学》2012年硕士论文　论文类型：学位论文

【摘要】：GML是基于XML对空间数据进行的编码，并为GIS领域和各企业广泛接受，是目前网络空间信息的重要组成部分。随着信息技术的发展，空间信息量急剧的增加；如何对GML空间数据进行有效的存储和管理成了影响GML发展的关键问题。依据GML空间数据自身的特性（地域分布性），使用集中式的存储不利于对空间数据的统一管理和访问；而云存储是一种商业化实现的分布式存储；它是将一些廉价计算机的空闲资源集中起来，共同分担海量数据的分布式存储和管理任务的。所以，对GML空间数据采用分布式存储和管理是一个比较好的方案。针对海量GML空间数据如何进行有效的存储，主要进行了以下几方面的研究及创新点分析：首先，需要对空间数据进行合理的划分，既要考虑空间数据多维的特性，，又要考虑其空间关系和拓扑关系等；对几种空间数据划分进行分析过后，又对Hilbert曲线层次分解算法和GML矢量图层分割的空间数据划分进行了研究，提出了对GML空间要素图层的划分并结合Hilbert曲线层次划分快速编码的方法对空间要素对象进行划分处理；在具体的应用中，参照分布式存储系统中各子节点的信息对要素对象进行分块划分，并分配相应子节点的空闲资源进行分布式存储；其次，在云计算环境下对分布式数据库的设计，结合GML空间数据的特征和Hadoop平台的主从式架构，对各主从服务器节点所存储的相关表结构、数据类型等进行了设计。针对分布式存储的GML空间数据的索引机制和索引结构表也进行了分析研究，主要研究了QR树的特性和对GML索引的构建；最后，在Hadoop开源平台的主从式架构下对分布式存储的原型系统进行了设计和部分实现，并通过实验对单机存储和分布式存储的效率及其查询访问的效率进行了比较分析，得出了对海量的GML空间数据无论是在存储还是查询访问，分布式计算机集群的处理效率都高于单机的处理效率。创新点主要有通过Hilbert曲线对GML空间数据采用要素队列的划分方法；结合hadoop平台对GML数据的并行空间索引机制的创建及各子节点索引结构表的设计；GML空间数据分布式存储流程的设计；云存储系统功能设计与实现等。
[Abstract]:GML is the XML of spatial data based on the encoding, and widely accepted as the GIS field and the enterprise, is an important part of spatial information network. With the development of information technology, the spatial information quantity increases sharply; how to effectively store and pipe of GML spatial data management has become the key issues affecting the development of GML. According to the characteristics of GML spatial data itself (regional distribution), the use of centralized storage is not conducive to the unified management and access of spatial data; and the cloud storage is a commercial implementation of distributed storage; it is free of some cheap computer resources together and share the distributed storage and management of massive data task. So, the spatial data of GML using distributed storage and management is a better solution.
In view of how to store the massive GML spatial data effectively, the following research and innovation are mainly carried out in the following aspects:
First of all, the need for a reasonable division of spatial data, it is necessary to consider the characteristics of spatial data cube, but also consider the spatial relations and topological relations; to analyze the data of several space division after Hilbert curve hierarchical decomposition of spatial data partitioning algorithm and GML layer were studied. The proposed division of GML the elements of space and layer method combined with the Hilbert curve hierarchy division fast encoding processing for space object elements; in the specific application, divided the elements of object for each sub node according to the information in the distributed storage system, distributed storage resources and free distribution of the corresponding child node;
Secondly, design and calculation of distributed database in the cloud environment, combined with the master-slave architecture of GML spatial data and the characteristics of Hadoop platform, the table structure of the master-slave server nodes stored, data types are designed. The distributed memory GML spatial data indexing mechanism and index structure of table analysis of the characteristics, mainly studies the QR tree and the GML index construction;
Finally, in the master-slave architecture of the Hadoop open-source platform under the prototype system of distributed storage has been designed and partially implemented, and make a comparison through experiments on the efficiency of single storage and distributed storage and query efficiency, the GML spatial data in the mass both in storage access and query efficiency of distributed processing the computer cluster is higher than single processing efficiency.
The main innovation division method by using factor queue for GML spatial data through the Hilbert curve; design and create sub node index structure table with the Hadoop platform of GML data parallel spatial indexing mechanism; design of GML spatial data distributed storage process; cloud storage and realize the functional design of the system.

【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP333

【参考文献】