Verifying Patterns of Dynamic Architectures using Model Checking
This addresses the risk of architectural flaws in software systems, which can lead to difficult-to-repair quality issues, though it is incremental as it applies existing model checking to pattern verification.
The paper tackled the problem of verifying architectural patterns for dynamic systems by proposing a model checking approach, discovering ambiguities and missing constraints in three well-known patterns (Singleton, Model-View-Controller, Broker).
Architecture patterns capture architectural design experience and provide abstract solutions to recurring architectural design problems. They consist of a description of component types and restrict component connection and activation. Therefore, they guarantee some desired properties for architectures employing the pattern. Unfortunately, most documented patterns do not provide a formal guarantee of whether their specification indeed leads to the desired guarantee. Failure in doing so, however, might lead to wrong architectures, i.e., architectures wrongly supposed to show certain desired properties. Since architectures, in general, have a high impact on the quality of the resulting system and architectural flaws are only difficult, if not to say impossible, to repair, this may lead to badly reparable quality issues in the resulting system. To address this problem, we propose an approach based on model checking to verify pattern specifications w.r.t. their guarantees. In the following we apply the approach to three well-known patterns for dynamic architectures: the Singleton, the Model-View-Controller, and the Broker pattern. Thereby, we discovered ambiguities and missing constraints for all three specifications. Thus, we conclude that verifying patterns of dynamic architectures using model checking is feasible and useful to discover ambiguities and flaws in pattern specifications.