ReSoFT: a reusable testbed for development and evaluation of software fault-tolerant systems

Abstract

The Reusable Software Fault-tolerance Testbed (ReSoFT) has been developed to facilitate the development and evaluation of high-assurance systems that require tolerance of both hardware and software faults. Central to ReSoFT is a library of reusable software components from which a wide variety of software fault tolerance (SWFT) techniques can be utilized to construct highly dependable systems. The reusable components include: (1) SWFT executive components, (2) SWFT support components, (3) SWFT interface components, (4) fault-tolerant network communication components, and (5) fault-injection components. A set of graphical tools are also provided to build, monitor and test the SWFT systems. The ReSoFT testbed is developed and hosted on a network of Sun workstations running the standard Solaris operating system. The workstations are connected with dual-redundant networks to tolerate communication faults. To ensure effective reuse, object-oriented analysis and design based on the Booch method have been used to develop the reusable components. The components were implemented in Ada 95 to take advantage of its new object-oriented and real-time support features. The graphical tools were implemented in Java which provides object-oriented and platform-independent features.