Layered architecture(s): principles and practice in concurrent and distributed systems

Abstract

Architecture plays a crucial role in the management of the development of complex systems, by identifying components and connections, and by showing how these may be aggregated or decomposed to give the various levels of detail essential for a complete understanding of what is intended to be developed or what has been developed. This paper sets out a form of layered architecture which can be used to describe a real-time distributed computing system from the functional, design, distribution and execution viewpoints. Strong traceability between these four layers is achieved by the common application of a structural concept based on independently operating processing components which interact with one another and with the system environment, through explicitly defined interaction components. Of particular importance is the set of protocols which characterise the interactions from a functional viewpoint, and which directly translate into route connections in a design model. Routes are shown to have well defined distribution and execution properties. Some implementation issues and application examples are discussed. The layered architecture approach grows out of and feeds into, real industrial applications. It is supported by development tools and generic components which have demonstrated their value in terms of economic development and product robustness.