Stability Metrics for Continuous Integration of Service-Oriented Systems

Abstract

One of the key principles of the service orientation is the standardised service contract. However, the assumption that the service contract is kept unmodified during the whole life-cycle of a system is not always held. Evolution changes on the service APIs have an impact on the maintainability of their programming clients within the system making difficult the continuous integration of the services. The metrics that have currently been applied to assess the service maintainability assess the service coupling, cohesion, complexity, and granularity. Software stability can further contribute in assessing the maintainability of systems. However, it is challenging to measure the stability of service APIs without having evolved their programming clients, because it should be measured by considering the types of the evolution changes in APIs that have direct impact on the programming clients. To address this challenge, we define a set of mappings between evolved service APIs based on which the stability changes can be determined. We further specify a generic algorithm that recognises the evolution changes required on the programming clients of the evolved APIs. We finally define an initial version of a suite of metrics that estimate the stability of a service system without assuming the existence of the evolved programming clients.