A method for selecting environments for software compatibility testing

Modern software is developed to work with multiple software and hardware architectures, to cooperate with various peer components and can be installed in many different configurations. In order to test it, all possible working environments needs to be created. This requires software and hardware resources like servers, networks and software licenses and most important: man-hours of qualified engineers that will have to configure and maintain them. Because resources are usually limited we have to choose a set of configurations with highest impact on quality of software under test. In this paper we present a method of measuring effectiveness of given software environment for discovering defects in software by introducing environment sensitivity measure. We also show how it can be used in simple algorithm used to select best configurations by using only a selected subset of them and progressively modifying it thougout software development process.