Theoretical basis for software fault tolerance
A new theoretical basis for software fault tolerance is developed. The theoretical basis has two parts: (1) a formal, set-theoretical model of checking and (2) an informal model of software execution. Central to the work presented is the notion of software fault localization, a notion overlooked in previous theoretical research. Analysis of the design of the recovery block and N-version programs reveals the serious flaw that neither performs software fault localization. This flaw has been carried into the many programming constructs which have grown out of the recovery block. The set-theoretical model of checking developed provides a conceptual framework within which run-time checks can be investigated, analyzed, and designed. Within the model, checking is considered abstractly in terms of relations similar to those found in relational data-base theory. Such relations provide an ideal representation of software specifications. The emphasis of the informal model of execution is upon run-time checking and data dependency in fault-tolerance software. The analysis of fault-tolerance software in terms of data dependency offers insights into the nature of state restoration and into the use of exception handling for providing software fault tolerance.
- Research Organization:
- Virginia Univ., Charlottesville, VA (USA)
- OSTI ID:
- 7230875
- Resource Relation:
- Other Information: Thesis (Ph. D.)
- Country of Publication:
- United States
- Language:
- English
Similar Records
An experiment in software fault elimination and fault tolerance
Structure design for fault tolerant distributed programs