Resilience, Survivability, and Elasticity: A Taxonomy for Change Impact Quantification of Reconfigurable Systems

Context. Modern distributed systems are flexible in moving from one configuration to another during operation in an automated or semi-automated manner, e.g., concerning dynamic CPU allocation and deploying updated versions of system services. Software architects need assurance that the system satisfies agreed quality of service (QoS) despite a change in system configuration. In the literature, under resilience, survivability, and elasticity, there are different change impact quantification approaches that each has different methods for quality metric specification, modeling a change, and impact analysis. However, independent of a particular approach, no taxonomy exists that clarifies a set of general concepts concerning change impact quantification in reconfigurable systems.Objective. We propose a taxonomy by examining existing approaches under the three meta-quality attributes for change impact quantification.Method. We start with works done by heavily cited authors behind resilience, survivability, and elasticity who provided unambiguous and measurable definitions.Result. We classify existing approaches for change impact quantification based on the taxonomy. We demonstrate the application of our taxonomy through an example.Conclusion. The taxonomy provides a unified and structured knowledge across communities that further eases communication and development of new approaches for change impact quantification.