基于Zabbix的云监控系统的设计与实现

发布时间：2018-06-26 05:53

本文选题：Zabbix + 云计算　；参考：《电子科技大学》2017年硕士论文

【摘要】：随着云计算时代的快速到来,现有的IT环境中存在着多种多样的设备,且其系统的构成也非常复杂,架构通常由上层应用、系统架构、底层支持构成,因此一款设备的健壮性、可靠性以及出现问题的及时响应都变得越发的重要。在设计与维护过程中,为了及时掌握整个系统的基础环境、业务应用流程、系统的可用性、通过获得的监控数据分析集群趋势,需要获取其中各个设备的运行状态,这一切都是离不开监控系统。目前市场上有较多优秀的开源系统,但是综合发现仅仅使用监控系统而不进行架构的重新设计以及添加相应所需功能,则无法跟上时代的潮流。监控系统的目的是监控集群里环境是否出现异常,但是目前市场上的开源产品没有保证其具有高可靠性以及监控系统本身出现问题的自动化处理。同时对所产生的日志没有进行很好的分析处理,大多数仅仅是将所监控数据从数据库中拿出显示在web页面上。如何设计出一款具有高可靠性、架构设计合理、可对监控日志进行分析的监控系统是本文研究的主要课题。本论文以基于Zabbix的云监控系统的设计与实现为主要内容,设计出高可靠性、结合Splunk日志分析处理的监控系统框架并实现。该监控系统的框架结合目前市场上三大主流开源监控产品(Ganglia、host-sflow、Nagios)的优点,以及吸取仿生物神经系统(BANS)的分层模式优点,仿生形成三层两域,三层自上而下为:服务端(类中枢神经层)、代理端(类周围神经层)、客户端(类神经元层);两域包括:类中枢神经域和类周围神经域。本云监控系统在此分层模式下,分别在程序、架构、数据库中提高本云监控系统的可靠性,并结合Splunk进行监控日志分析处理,与此同时,可基于Zabbix设计出具有防误报的、梯度的报警功能。根据主流监控系统的优化思想,在通信协议的选择中,与实际生产环境相结合,在同一个IDC中通信的客户端插件间通信协议采用UDP,不同的IDC机房的客户端插件间通信协议采取TCP,这样在保证可靠性的同时优化了资源的占用;在数据库的选择与处理上,采用mysql数据库并采用Mysql Cluster技术,且对数据库进行优化,以解决监控数据过多而造成的数据库瓶颈和保证数据库的可靠性。最终实现基于Zabbix的CloudMonitor云监控系统。
[Abstract]:With the rapid arrival of cloud computing era, there are a variety of devices in the existing IT environment, and the composition of its system is very complex. The architecture is usually composed of upper application, system architecture, and bottom support, so a device is robust. Reliability and timely response to problems have become increasingly important. In the process of design and maintenance, in order to grasp the basic environment of the whole system, the business application flow, the usability of the system in time, we need to obtain the running state of each equipment through analyzing the cluster trend through the monitoring data obtained. All this is inseparable from the monitoring system. At present, there are many excellent open source systems in the market, but it can not keep up with the trend of the times if we only use the monitoring system and do not redesign the architecture and add the corresponding functions. The purpose of the monitoring system is to monitor whether there are any anomalies in the environment of the cluster, but the open source products in the market do not guarantee its high reliability and the automatic handling of the problems in the monitoring system itself. At the same time, the generated log is not well analyzed, most of the monitoring data is only taken out from the database and displayed on the web page. How to design a monitoring system with high reliability, reasonable architecture and analysis of monitoring log is the main subject of this paper. In this paper, the design and implementation of the cloud monitoring system based on Zabbix is the main content, and a high reliability monitoring system framework is designed and implemented in combination with Splunk log analysis and processing. The framework of the monitoring system combines the advantages of the three major open source surveillance products (Gangliahost-sflow Nagios) in the market at present, as well as the advantages of the hierarchical model of the biomimetic nervous system (Bans), and bionics forms three layers and two domains. The three layers are: serous (central nervous layer), agent (peripheral nerve layer), client (neuron-like layer), and the two domains include: central nervous region and peripheral nerve domain. In this hierarchical mode, the cloud monitoring system improves the reliability of the cloud monitoring system in the program, structure and database, and carries out monitoring log analysis and processing with Splunk. At the same time, it can be designed based on Zabbix to prevent false positives. Gradient alarm function. According to the optimization idea of mainstream monitoring system, in the choice of communication protocol, it combines with the actual production environment. In the same IDC, the communication protocol between clients and plug-ins adopts UDP.The communication protocol between clients in different IDC rooms adopts TCP, which not only ensures reliability, but also optimizes the usage of resources. Mysql database and MySQL Cluster technology are adopted to solve the bottleneck of database caused by too much monitoring data and to ensure the reliability of database. Finally, the cloud monitor system based on Zabbix is implemented.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP277

【参考文献】