Research And Implementation Of The Enterprise-level Application Monitoring System

Internet is changing the whole world everywhere, but also changes the global commercial operation mode. The rapid development of enterprise information system is also showing a pretty trend of fast growth. A world enterprise maybe can have more than hundreds of Application systems. And in the actual these application systems also take more and more important role in the daily life. Many enterprises IT system has also implemented the service window for 24 hours and 7 days. But the system is unable to run healthily for all the time, maybe one day it will crash. So it is a very important thing for supporting engineer to figure it out when the IT application system is crashed. For the supporting engineer, they need to know the whole of the application systems when they are running. They need to know whether the CPU and Memory of the hosts are high or not, whether the network flow is too large or not, if there will be the risk of clogging; whether the application process is existing or not, working or not; whether the middleware containers are able to supporting more business activities or not; whether the application systems have some bugs hidden behind; whether there will be risk triggering a major fault. Therefore, we need to know all of the information when the application systems are running. If we find there are some problems where the application systems are running, we can immediately do something to solve the problems and recover the systems. This study focuses on the enterprise level application monitoring system, is different from many other free open monitoring systems which are just only focusing on monitoring physical hardware level, such as the host layer, network layer, and not to monitor the application system.The application monitoring principles proposed in this paper is based on the careful analysis and does the summary for the present application system architecture and business processes, throughout the area coverage of the enterprise application system of each index in the daily operation, and can alarm in the first time in a variety of ways.It has the following characteristics:In order to achieve the real-time monitoring for all of the enterprise-level application systems for 7days*24hours. Locally it uses the very cheap and effective Haproxy+keepalived architecture, to provide the system load balancing, at the same time, the application monitoring system is deployed in other different places, to improve the monitoring system availability.By using agents-server architecture, monitoring system server provides only receiving monitoring information and send monitoring alerts, regardless of whether the enterprise application system is established on the Linux host, UNIX host or Windows host.In order to ensure the different priority monitoring can have the higher priority, the application monitoring system run the high priority monitoring first and run the low priority monitoring lately. The very importing key functions in the system are run and told firstFor all the hundreds of business systems in enterprises, enterprise monitoring system ensure that the monitoring systems are independent for all systems, and are not affected by each other.It has the following advantages:The monitoring system covers infrastructure layer, Web layer, application layer and business logic layer, much far better than the open source systems currently on the market which only provider the physical layer monitoring.The monitoring system provides lots of ways to send the alert messages, including the phone, SMS, APP push, mail, page tips and so on, to ensure that application maintenance engineer can receive the alert message in the first time.By using agent-server architecture, monitoring system server only receives the monitoring information and sent the monitoring alarm messages.