海量地震数据并行处理系统架构优化设计与实现  

Optimization Design and Implementation of Parallel Processing System Architecture for Massive Seismic Data

作　　者：李书平 刘树仁[1,2] 朱启伟 王西林[1,2] 

机构地区：[1]中国石油勘探开发研究院西北分院 [2]中国石油天然气集团有限公司物联网重点实验室

出　　处：《信息技术与标准化》2024年第4期96-99,共4页Information Technology & Standardization

摘　　要：随着石油勘探数据规模成指数级增长,各类应用软件系统的数据计算量和磁盘存储量也随之增加,给传统并行处理架构带来巨大挑战。提出了海量地震资料并行处理系统架构的设计与优化方案。通过分析地震资料处理过程中各软件模块对CPU、内存、本地磁盘及网络传输等硬件参数的敏感性和NFS、RPC、Buffer调优配置,实现了该集群的高效运行。实际生产应用表明,该在支持百TB级海量数据并行处理、多操作系统软件兼容问题等方面有着独特优势,有力支撑了地震资料处理业务发展。With the exponential growth of petroleum exploration data,the data computation and disk storage of various application software systems have also increased,posing great challenges to traditional parallel processing architectures.This paper proposes the design and optimization scheme of a parallel processing system architecture for massive seismic data.By analyzing the sensitivity of each software module to hardware parameters such as CPU,memory,local disk and network transmission during seismic data processing,as well as optimizing the configuration of NFS,RPC and Buffer,the efficient operation of the cluster has been achieved.Actual production applications have shown that the system has unique advantages in supporting parallel processing of massive data at the level of 100 TB,as well as compatibility issues with multiple operating system software,which strongly supports the development of seismic data processing business.

关 键 词：网络文件系统 远程过程调用 缓冲区 NAS存储 CPU 

分 类 号：P631.44[天文地球—地质矿产勘探] TP274[天文地球—地质学] P618.13[自动化与计算机技术—检测技术与自动化装置]