NetCDF4 性能分析与调优研究-《计算机技术与发展》

文章信息/Info

Author(s):: YANG Yu-hong; KOU Yuan-yuan; QIAO Wen-wen; Guangdong Province Meteorological Data Center,Guangzhou 510000,China

Keywords:: NetCDF4; numerical prediction; storage efficiency; performance research

摘要:: 介绍了 NetCDF4 在气象格点数据服务系统中的应用,对比了同种数值预报模式使用不同分块策略时创建、更新文件时间,以及在不同压缩系数时的性能。通过性能分析了解了 NetCDF4 的新特性,进行了性能调校,并在此基础上总结出了 NetCDF4 的调优策略。合适的调优策略会显著减少文件的创建(更新)时间,或提高文件的压缩率,节省系统的存储空间。该研究结果为 NetCDF4 在数值预报业务中的全面使用奠定了基础,为提高数值预报业务存储效率提供了可靠的实践依据。 NetCDF4 合理适当的使用将会显著提高气象格点数据服务系统的服务效率,也是未来几年“广东省气象格点数据服务系统冶的主要工作方向。

Abstract:: The application of NetCDF4 in the meteorological grid data service system is introduced,and the performance of the same numerical prediction model when creating and updating files using different blocking strategies,and that at different compression factors are compared. Through performance analysis,the new characteristics of NetCDF4 are understood,the performance is adjusted,and on this basis,the optimization strategy of NetCDF4 is summarized. A suitable tuning strategy can significantly reduce the file creation (update) time,or increase the file compression rate,saving system storage space. The research lays the foundation for the comprehensive use of NetCDF4 in the numerical forecasting business,and provides a reliable practical basis for improving the storage efficiency of numerical forecasting services. The reasonable and appropriate use of NetCDF4 will significantly improve the service efficiency of the weather grid data service system,which is also the main working direction of the “Guangdong Province meteorological grid data service system” in the next few years.