ARX-Based Aging Modeling of Application Server and System Rejuvenation Scheme

In order to study the performance decline of application servers, this paper designs the customer requesting procedures and server monitoring programs for multi-scenes, which records five types of parameters, and the dimensionality reduction is implemented by using the Principal Component Analysis (PCA) firstly, then the multidimensional time series analysis model for the key performance indexes of application server middle-ware is built in this paper. The paper also proposes a kind of system rejuvenation scheme according to the fitting model. Measuring data and analysis results show that the proposed software aging model can better fit raw data, and predicting results can be used as the trigger basis of system rejuvenation action.