面向大规模集群作业并发规模的数据库连接池优化技术  

Application of database pool technology for job concurrent scale in large-scale cluster environment

作　　者：师伟 王向辉 林茂春 侯红军 程实 SHI Wei;WANG Xianghui;LIN Maochun;HOU Hongjun;CHENG Shi(Bureau of Geophysical Prospecting INC.,China National Petroleum Corporation(BGP),Zhuozhou 072751,China)

机构地区：[1]中国石油东方地球物理公司,涿州072751

出　　处：《物探化探计算技术》2024年第2期235-241,共7页Computing Techniques For Geophysical and Geochemical Exploration

基　　金：中国石油集团科学研究与技术开发项目(2021ZG03)。

摘　　要：数据库作为地震勘探处理软件系统的多学科数据存储管理的核心组件,在底层支撑着处理作业的读写访问需求。在当前,随着地震勘探处理技术的飞速发展,大规模集群下处理作业的并发规模也快速扩张,底层数据库采用常规的读写一体化部署方式难以支撑大规模并发作业的读写请求。笔者针对地震勘探大规模集群数据资料处理作业并发场景,提出一种“1+N”读写分离部署方式的数据库连接池优化技术,设计了基于数据库服务器节点信息的资源分配器,对并发作业的数据库读写请求进行了合理的优化,并在实验室环境和实际地震勘探数据资料处理生产中进行了验证,能够支撑大规模集群下处理作业长事物、高并发等特征的数据库访问请求。As the core component of multidisciplinary data storage and management of seismic exploration processing software system,the database supports the reading and writing access requirements of processing jobs at the bottom.With the rapid development of seismic exploration processing technology,the concurrent scale of processing jobs under large-scale clusters is also expanding rapidly.The e's conventional read-write integrated deployment mode makes it challenging to support the read-write requests of large-scale concurrent jobs.This paper presents a concurrent scenario for data processing of large-scale cluster seismic exploration,The database adopts“1+N”The connection pool optimization technology of“read-write separation deployment mode”designs a resource allocator based on the node information of the database server,reasonably optimizes the database read-write requests of concurrent jobs,and has been verified in the laboratory environment and actual seismic exploration data processing production.It can support processing long things and high concurrency under large-scale cluster Database access requests.

关 键 词：数据库 读写请求 集群作业 并发规模 

分 类 号：P631.4[天文地球—地质矿产勘探]