Towards a model-driven method for reliable applications
Vidar Slåtten, Frank Alexander Kraemer, Peter Herrmann · 2010
We present a model-driven method to incrementally introduce fault-tolerance mechanisms into application models that are initially developed with assumptions of ideal transmission semantics. As main structuring units, our models use collaborative building blocks in UML that can encapsulate the behaviour of several participants in order to perform a certain task. Since these building blocks can be designed and analysed separately, fault-tolerance mechanisms can be introduced block by block, which reduces the size and complexity of specifications that have to be understood at a time. Applying fault tolerance at the application layer also brings the benefits of easily porting applications to other platforms and applying model-level analysis tools to the fault-tolerance mechanisms themselves. We illustrate our method through the development of an access control system.