Abstract
Software fault tolerance is often necessary, but can itself be dangerously error-prone because of the additional effort that must be involved in the programming process. The additional redundancy may increase size and complexity and thus adversely affect software reliability. Object-oriented programming provides an appropriate framework for controlling complexity and enforcing reliability. However, software fault tolerance cannot be achieved merely by implementing the classical fault-tolerance schemes in an object-oriented fashion. New problems arise while integrating software redundancy into object-oriented computing systems. This paper identifies a set of such problems, addresses possible solutions, and proposes an object-oriented architecture for dealing with software design faults. Both linguistic supports for the architecture and implementation issues are discussed in detail.
Toward an Object-Oriented Approach to Software Fault Tolerance
In Fault-Tolerant Parallel and Distributed Systems,
Avresky, D.R. (ed), IEEE Press, 1994, ISBN 0-7923-8069-X
[Abstract]