In distributed systems the communication paradigm used for intertask interaction is the message exchange. Several approaches have been proposed that allow the specification of the dataflow between tasks, but in realtime systems a more accurate definition of these data flows is mandatory. Namely, the specification of the required tasks' and messages' parameters and the derivation of the unspecified parameters have to be possible. Such an approach could allow an automatic scheduling and dispatching of tasks and messages. The data streams present a possible approach to the holistic scheduling and dispatching in real-time distributed systems where different types of analysis that correlate the various parameters are done. The results can be used to fulfil different goals like the level of buffering that is required at each node of the distributed system