Distributed real-time applications share a group of processors connected by some local area network. A rigorous and sound methodology to design real-time systems from independently designed distributed real-time applications is needed. In this paper, we study a distributed real-time system design scheme using CBS-based end-to-end scheduling. The scheduling scheme utilizes CBS to allocate both CPU shares and network bandwidth to a distributed real-time application when it arrives at the system. Our proposed solution uses the same scheduling paradigm for both resources. In this way, we believe the system can have a more consistent scheduling objective and may achieve a tighter schedulability condition.