| With the rapid development of the Internet,monitoring alarm system is one of the core modules in large-scale operation management system.Monitoring alarm system not only can realize the real-time monitoring and alarming of equipment’s performance and information security,meanwhile,it can also make statistical analysis that relevant to the behaviors of security incidents and security alarms,indeed,it can realize the real-time prediction about the trend of security incidents and real-time warning of security alarm action.However,there is one shortcoming in the current monitoringsystem.The shortcoming is that in the process of hierarchy data transmission,data fault tolerance and real-time monitoring need to be imporved.Based on duplicate hot standby technology,this thesis proposes a distributed data fault-tolerant scheme,then based on this method,with NoSQL and Ganglia,we realize a monitor system.The system can monitor system performance,data flow,security incidents and security alarm behaviors,then it can give real-time alarming.This monitoring system adopts non-relational database to realize distributed cluster storage,which can reduce the machine’s pressure due to rapid query operations and improve the stability of the storage module.With regular expressions,log parsing can enhance the accuracy of required data and reduce garbage data.Meanwhile,the system has a good scalability,only simply adding machines,then it can achieve duplicate hot standby and get a good extension.The system not only can show the current network’s status,but also can record,analyze and display the latest log of network devices in a cluster.Finally,we make a verification test of the monitoring system.Experimental results show that the monitoring system,which based on the scheme we proposed,can make a good real-time monitoring and alarming,deal with abnormal conditions,which includes network fault and downtime,and the monitoring system have a good data fault tolerance and stability. |
PDF Full Text Request