Development of Fault-Tolerant Software Systems Based on Architectural Abstractions

The incorporation of fault tolerance into systems normally increases their complexity, which consequently makes their analysis more difficult. This paper discusses how architectural abstractions can be effective in developing fault-tolerant software systems. Depending on the fault model and the resources available, different abstractions can be employed for representing issues that are related to fault tolerance, such as error detection, and error and fault handling. These architectural abstractions, and their internal views, can be instantiated into concrete components and connectors for designing fault-tolerant software architectures. Since structural and behavioural properties associated with these abstractions are formally specified, the process of verifying and validating software architectures can be automated. In this paper, we show how appropriate architectural abstractions and a recursive process can facilitate the architectural modelling and analysis of fault-tolerant software systems. The feasibility of the proposed approach is demonstrated in the context of a critical real-time application.