A method for software aging state evaluation

The software aging modeling and state evaluation method of a real VOD (Video on Demand) system are investigated. Firstly, the data on several system resource usage and application server are collected. Then, the linear regression method is adopted to remove the various workloads' affection on the system resource consumption, and the software aging process is obtained. Next, an information entropy-based iterative self-organizing data analysis technique arithmetic (IE-ISODATA) is proposed, the statistical metric of a segment is calculated by the information entropy, and the distance between two different segments is evaluated by the relative entropy. Based on the IE-ISODATA method, the software system aging state under different interval is obtained. Finally, several experiments are taken to evaluate the proposed methods. The results show that the abnormal consumption of system resource reveals the software aging process after the analysis of linear regression model. And the software aging states are gotten by using the presented IE-ISODATA method with better effect and high efficiency. Based on the models employed here, software rejuvenation policies can be triggered by actual measurements.