Fundamental concepts for fault tolerant systems

Abstract

In order to be able to think clearly about any subject we need precise definitions of its basic terminology and concepts. If one reads the literature describing fault tolerant computing there is less agreement on fundamental models, concepts and terminology that would perhaps be expected. There are well established usages in particular subcommunities and many other individual workers take care to use terms carefully. Unfortunately there are also many papers in which terms are freely applied to concepts in an arbitrary and inconsistent way.

This thesis attempts to bring together some of the concepts of fault tolerant computing and place them in a formal framework. The approach taken is to develop formal models of system structure and behaviour, and to define the basic concepts and terminology in terms of those models. The model of system structure is based on directed graphs and the model of behaviour is based on trace theory