In this paper, we discuss the experience we gained during three performance engineering projects that we did in co-operation with telecommunication industry. In each of the projects, we analyzed the performance of a real-time system, which has large embedded software. Each system was at a different stage of development. We used three different modeling techniques: Queueing networks, execution graphs, and message sequence charts. We also applied different approaches of building models. To give guidelines for analyzing large systems, we discuss our method of analysis, rationale for choices we made, and the lessons we learned.
展开▼