In complex hard real-time systems with tight constraints on system resources, small changes in one component of a system can cause a cascade of adverse effects on other parts of the system. We address the inherent complexity of making architectural decisions by raising the level of abstraction at which the analysis is performed. Our analysis approach gives the system architect a rigorous method for quickly determining which system architectures should be pursued, and it allows the architect to track and manage the cascading effects of subsystem/component changes in a comprehensive, quantitative manner. The end product is a virtual architecture analysis that systematically incorporates the inherent coupling among interacting system components that share limited system resources.