随着地理信息存储量的飞速增长,传统的单进程、集中式的数据处理方式已不能满足基于网络的地理信息服务的效能要求。分析对比了OpenMP,MPI和MapReduce等主流并行编程模式,将关系型数据库与分布式空间数据管理系统相结合,提出了面向并行处理的地理信息存储模型和数据组织模型,将该模型与传统模型进行了对比分析,并基于MapReduce实现了地理空间数据并行处理框架,选取了矢量数据装载、影像数据装载以及数据切片作为典型数据处理案例开展对比实验,该技术方案的处理效率均数倍于传统技术方案。实验表明,该模型能够很好地支持并行处理框架,可为分布式环境下数据处理中心构建提供一个有效解决方案。
With the rapid growth of geographic information storage, traditional single process and centralized data processing could not satisfy the requirement of the web-based geographic information service. OpenMP, MPI, Ma- pReduce and other mainstream parallel programming mode were analyzed and compared, and the parallel process- ing oriented geographic information storage model and data organization model were presented by combining rela- tional database with distributed spatial data management system. Through comparing this model with the traditional model, geospatial data parallel processing framework based on MapReduce was implemented, and the loaded vec- tor data, image data and data slice were selected as typical case of data processing to carry out comparative experi- ments. The processing efficiency of the technical solution was many times greater than traditional technical pro- grams. Experiments show that this model can support parallel processing framework and provide an effective solu- tion for building data processing centers in the distributed environment.