Layered Modeling of Hardware and Software, with Application to a LAN Extension Router

Understanding the interactions between hardware and software is important to performance in many systems found in data communications like routers. Responsibilities that traditionally were programmed in software are being transferred to intelligent devices, and special purpose hardware. With more functionality being transferred to these devices, it becomes increasingly important to capture them in performance models. Modeling hardware/software systems requires an extended queueing model like LQN. This paper describes a layered architecture model which represents hardware and software uniformly and which emphasizes resources and performance, called a Resource-based Model Architecture (RMA). The approach is demonstrated on a remote access or LAN extension router. The model is created by a systematic tracing of scenarios and is used to explore the router capacity for different workloads, and to analyze a re-design for scaleup.