基于MapReduce的交互可视化平台

摘要

针对长时间序列的海洋环境信息交互可视化处理问题,本文提出一种基于云环境的交互可视化平台架构.可视化平台以Hadoop为基础,将GPU 、MPI并行计算引入MapReduce处理机制,实现海洋环境空间数据场大规模数据集的检索、抽取、插值计算、特征可视化分析的并行处理,达到海量数据的远程交互可视化处理目的.实验结果表明,本文提出的可视化平台架构可以有效地将MPI、GPU并行计算与MapReduce处理机制相结合,以满足海洋环境信息交互可视化的需求.%For dealing with the problem of ocean data's interactive visualization, this paper proposes an interactive visualization platform architecture based on cloud computing. It inserts GPU, MPI parallel computing into MapReduce mechanism of Hadoop to realize the parallel processing of large-scale ocean environment data sets, such as data retrieval, data extraction, data interpolation, analysis of characteristics' visualization, so that the massive data' s visualization can be processed in the remote interactive way. The experiment results show that the visualization platform architecture can combine MPI, GPU with MapReduce effectively to meet the needs of ocean data's interactive visualization.