Formal guides for experimentally verifying complex software-implemented fault tolerance mechanisms

Describes a framework allowing the experimental verification of complex software-implemented fault-tolerance algorithms and mechanisms (FTAMs). This framework takes into account two of the most important aspects which are increasingly required in newly-developed fault-tolerant systems: the considerations of COTS (commercial off-the-shelf) based architectures and the compliance with severe safety certification procedures. The strategy proposed shows how a rigorous FTAM specification, based on a multiple-viewpoint architectural description, may help to mechanically monitor the verification of its implementation under real conditions. The proposed strategy has been instantiated using two mechanized techniques: model checking and fault injection. The preliminary conclusions of the application of this automated approach to a small part of a commercial fault-tolerant system help us clarify its usage and its suitability for validating complex dependable systems.