Combining Static and Dynamic Impact Analysis for Large-Scale Enterprise Systems

Software changes and their impact on large-scale enterprise systems are critical, hard to identify and calculate. A typical enterprise system may consist of hundreds of thousands of classes and methods. Thus it is extremely costly and difficult to apply conventional testing techniques to such a system. In our previous work [1], a conservative static analysis with the capability of dealing with inheritance was conducted on an enterprise system and associated changes to obtain all the potential impacts. However, since static analysis takes into account all the possible system behaviours, the analysis often results in a good number of false-positives and thus over-estimation of the impact on other methods in the system. This work focuses on extending our previous static approach by an aspect-based dynamic analysis, to instrument the system and collect a set of dynamic impacts at run-time. The new approach is still safe, but more precise than the static analysis. Safety is preserved since the static analysis serves as the input source to the dynamic analysis, and we are careful not to discard impacts unless we can show that they are definitely not impacted by the change. It is more precise since dynamic analysis examines behaviours that do definitely occur at run-time and hence is able to reflect the real impacts. Additionally, our analysis is able to handle the scalability issue. The targeted system is orders of magnitude larger than the system other existing approaches can deal with. A case study was conducted to illustrate that specific objectives can be attained.