A General Noninvasive Application Monitoring Systemfor Service Invocation TopologyAnalysis And PerformanceBased On Docker

Most enterprise applications are complex distributed applications, they are composed of multiple services which may be developed by different teams with different languages and are distributed on the different locations in the network. The situation makes these applications difficult to debug and determine the bottleneck of system, it is even more difficult to quickly locate the problem when exceptions happen. with the development of the Docker, more and more enterprise applications have been deployed in Docker cluster, which puts forward a new challenge for the monitoring of applications. And the monitor system of Docker is mainly used for monitoring the usage of platform resources as well as the state of Docker containers, but it can not monitor the overall performance of applications, once an exception occurs or performance reduces, Docker cluster could do nothing.For traditional distributed applications, there have been some methods which can provide the corresponding monitoring function, however, these methods may intrude applications or platforms, or require users to provide relevant information, or can not provide real-time monitoring analysis, or can only be adapted to some pecific kind of applications. In short, traditional methods are not suitable for overall performance monitoring for distributed applications in Docker cluster.Therefore, in this paper, we designed and implemented a general noninvasive monitoring system, Topoloror, for service invocation topology analysis and performance for applications in Docker cluster. Topoloror analysis topology relations created by invocations among services and performance of services. Topoloror may not obviously increase the load of applications and Docker cluster, but provide real-time monitoring service for all kinds of distributed applications in Docker cluster and provide users with visual analysis results of invocation topological relationship among services and performance of services, and provides the alarm when something unexcepted happens.