System level performance analysis methods play an important role in the design process of complex embedded systems. They allow to analyze essential performance characteristics of a system design in an early design phase and consequently support the choice of important design decisions before much time and resources are invested in detailed implementations.While formal analysis based methods for system level performance analysis lead to hard bounded analysis results and can thus be employed in the design of hard real-time systems, these methods are often restricted in their modeling and analysis capabilities, and the obtained results are often overly pessimistic due to a lack of details such analytical methods can incorporate in their system analysis.In this thesis we identify challenges for system level performance analysis of embedded systems, and based on these challenges we further develop and extend a framework for formal modular performance analysis and design of complex distributed embedded real-time systems.