Dynamic structure in software architectures

Abstract

Much of the recent work on Architecture Description Languages (ADL) has concentrated on specifying organisations of components and connectors which are static. When the ADL specification is used to drive system construction, then the structure of the resulting system in terms of its component instances and their interconnection is fixed. This paper examines ADL features which permit the description of dynamic software architectures in which the organisation of components and connectors may change during system execution.The paper outlines examples of language features which support dynamic structure. These examples are taken from Darwin, a language used to describe distributed system structure. An operational semantics for these features is presented in the π-calculus, together with a discussion of their advantages and limitations. The paper discusses some general approaches to dynamic architecture description suggested by these examples.