A Nonlinear Approach to Modeling of Software Aging in a Web Server

Software aging refers to the phenomenon that the performance degradation, increasing failure rate or massive resource usage demonstrated in long-running software systems such as web servers. It was reported that software aging has a significant relationship with resource usage. To study the software aging phenomenon in web servers, we built an experimental platform with a server and multiple clients to collect the data of system resource usage on the server. In this paper, we present several experiments based on the platform and study software aging phenomenon of a widely used web server software system - the Apache HTTP server. Some repeatable software aging phenomena are observed in our experiments. The process of software aging is shown to be nonlinear and chaotic. In order to better present the nonlinear interplay of various resource parameters of web server and forecast resource exhaustion, a nonlinear dynamic model of software aging is constructed to formulate software aging process.