JMeter-based aging simulation of computing system

In order to study the laws of the decline performance and determine the main reason of performance degradation, we use simulation methods to build simulation environment of the recession computing systems. Considering the impact of various unexpected factors, in this paper we use the JMeter-based distributed test structure and point-to-point test structure, combining with self-developed system resources monitor to set up a computing system simulation environment, reproducing the process of recession, and ultimately identifying the possible anti-granularity. In experiment we periodically send load through controlling the client. And we restart different application services by controlling the server-side respectively. We analyze the data of system resources that is obtained by these two different experimental programs, and then identifying the main reason that leads to the decline of performance is memory loss. At last, through the release conditions of different applications of the system resources, we conclude that major application service that leads to computing systems memory loss is the Tomcat server. The experimental results can provide data support for developing an appropriate performance evaluation indicators and fine-grained software rejuvenation strategy.