Real-time networks are used by distributed embedded systems of increasing complexity. This leads to a need for precise design and configuration techniques for these networks. This paper presents an approach for the iterative configuration of a real-time network and the associated distributed real-time control application that is to be deployed on it. The approach consists in combining network modeling and Component-Based Software Engineering techniques to automatically calculate the model of the data streams that will be exchanged over the network. The paper explains how to calculate the parameters used in data stream models from the specification of the application architecture and behavior. A single specification is used to generate both the data stream models and the infrastructure code of the application, thus providing guarantee of consistency between the behavior of the actual system and the network configuration. The use of this approach is demonstrated on an example which uses a TSN network.