A Methodology for Making Performance-Based Comparisons with Architectural Information

This paper demonstrates an approach to making early life cycle comparative performance assessments of alternative system architectures. Early problem identification reduces the likelihood of implementing designs that fail to meet performance goals. We restrict our technique to estimating performance differences early in the development phase, which is a critical step in identifying performance limitations. Designers routinely make assumptions both about the performance of the hardware platform and the software that will be used. Our approach simulates a queuing network model to provide a basis for making comparative performance assessments of architectures considered for system implementation, incorporating relevant known performance estimates. We will present results from the application of this technique to a case study that includes two ALOHA satellite network architectures. In this case, our early lifecycle assessment technique achieved a high degree of agreement with similar theoretical performance analysis models. The technique has the potential to allow a designer to investigate more easily "what-if" questions, enabling them to better understand the performance implications of alternative architectural proposals