Software Aging Detection Based on Differential Analysis: An Experimental Study
In this study we evaluate the applicability of the differential software analysis approach to detect memory leaks under a real workload. For this purpose, we used three different versions of a widely used software application, where one version was used as baseline (memory leak free) and the other two subsequent versions were our research subjects, one of them is confirmed to suffer from memory leaks and the other is memory leak free. The latter two versions were used to evaluate the accuracy of the proposed approach to detect aging, with respect to the number of true and false alarms. The results confirmed the previous findings obtained with synthetic workloads. The heap usage was a better metric than the resident set size, which has been extensively used for detecting software aging related to memory leakage. Also, the best data processing techniques to combine with the heap usage metric were Cumulative sum control chart and exponentially weighted moving average.
