一种大规模MongoDB集群监控方案的设计与实现  

Design and Implementation of a Monitoring Solution for Large-Scale MongoDB Clusters

作　　者：李云婷 张海明[1] 黎建辉[1] Li Yunting;Zhang Haiming;Li Jianhui(Computer Network Information Center,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院计算机网络信息中心,北京100190 [2]中国科学院大学,北京100049

出　　处：《科研信息化技术与应用》2019年第4期30-40,共11页E-science Technology & Application

基　　金：中国科学院战略性先导科技专项(A类)子课题“大数据资源库与门户系统”(XDA19020104)

摘　　要：如今诸多应用趋于使用MongoDB进行数据存储与管理,鉴于MongoDB集群规模庞大、结构易变动,研究如何实现针对大规模分布式MongoDB集群的实时性能和健康状态监控,具有重要技术挑战,对后续集群性能分析优化与故障恢复等具有重要意义。本文提出了一种集群结构描述方法和分布式监控架构,将集群实例作为基本监控单元,突破集群节点的局限性,实现监控数据统一存储,按照集群节点组织完成性能展示和分析,并通过开源平台Zabbix进行实现。实践表明系统能够灵活、简便且高效地监控大规模MongoDB集群,满足实时性能和健康状态监控要求。Nowadays,plenty of applications tend to use MongoDB for data storage and management.Given the largescale and changeable structure of MongoDB clusters,how to monitor real-time performance and runningstatus of large-scale distributed MongoDB clusters really becomes a technical challenge,which is also of great significance for cluster performance optimization and fault recovery.This paper proposes a method todescribe the structure of a cluster and a distributed monitoring architecture.In the architecture,the clusterinstance is used as the basic monitoring unit to break through the limitations of cluster nodes.Moreover,the monitoring data is stored in a unified manner and the performance display is organized according tothe cluster nodes.With Zabbix implemented,it’s proved that the system can monitor MongoDB clustersflexibly,simply and efficiently and meet the real-time performance monitoring requirements.

关 键 词：监控 MongoDB集群 分布式 Zabbix 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]