A reliable mobile agents architecture

Abstract

This paper describes the design of a novel mobile agent system that supports the flexible and reliable interaction of autonomous components in an object-oriented distributed system. It discusses the object-oriented design of the overall system together with a number of important components in terms of the most important design decisions. A detailed description of the reliability model is given in terms of the failure hypothesis and the related recovery protocols. Special emphasis is given to the generality and efficiency of the implementation and a number of preliminary experiences are described.