Analysis of a deployed software

Abstract

Analyzing a deployed software provides a means to characterize and leverage the software's runtime behavior as it is employed by its intended users. Preliminary studies have shown that leveraging the information obtained from the field provides engineers an opportunity to improve their software testing activities. The analysis of a deployed software can be performed in three stages: (1) the analysis to determine, before the software is deployed, where the instrumentation probes should be inserted into the software and what information that they should capture, (2) the analysis to determine when the field data should be sent back to the company during deployment, and (3) the analysis to leverage the field information after deployment. To make the analysis activities more feasible, we need to take into consideration that there are distinct characteristic differences between the development and the deployed environment. Deployed environment allows for less overhead,provides less control for the engineers, and requires highlyscalable techniques due to the high volume of information. Hence, the existing approaches for in-house analysis may become ineffective, inefficient, or even useless when they are directly applied to the deployed environment. Existing approaches for analyzing deployed software also need to be more aware that a technique in one analysis stage may affect the performance of a technique in other analysis stage. This research proposal details the challenges that arise when analyzing a deployed software and seeks to develop a set techniques to address these challenges that can be applied to each stage or across the analysis stages.