A model for adaptive fault-tolerant systems

Abstract

An adaptive computing system is one that modifies its behavior based on changes in the environment. Since one common type of environment change in a distributed system is network or processor failure, fault-tolerant distributed systems can be viewed as an important subclass of adaptive systems. As such, use of adaptive methods for dealing with failures in this context has the same potential advantages of improved efficiency and structural simplicity as for adaptive systems in general. This paper describes a model for adaptive systems that can be applied in many failure scenarios arising in distributed systems. This model divides the adaptation process into three different phases-change detection, agreement, and action-that can be used as a common means for describing various fault-tolerance algorithms such as reliable transmission and membership protocols. This serves not only to clarify the logical structure and relationship of such algorithms, but also to provide a unifying implementation framework. Several adaptive fault-tolerant protocols are given as examples. A technique for implementing the model in a distributed system using an event-driven approach for composing protocols in parallel is also presented.