9804113 This research addresses two important problems in the design of large and complex distributed software systems, namely: a) the need to increase the reusability of software components by reducing, or in some cases, eliminating their dependency on interconnection patterns, and b) the need to establish confidence that the final architecture will meet performance requirements by the proper selection of interconnection patterns between atomic components and aggregate components. More specifically, the project focuses on the design of a language to describe the architectural and performance characteristics of components, connectors, and patterns of interconnected components as architectural building blocks; the composition of patterns and target system architectures from the building blocks; and constraints on the composition of patterns and target system architectures. This research also addresses developing a method for mapping a target system to an analytic performance model used to predict the performance of the target systems under various workload intensity situations. By characterizing the performance of components and their interactions, the overall performance of the target system can be estimated. The biggest impact of this research is in advancing the methods and software technology needed to reduce the costs and risks in the software development of large and complex distributed software systems.***
