Configuration management in distributed real-time systems and applications is considered. In particular, an idea for specifying the SW and HW architectures and modeling the dynamical behaviour of embedded real-time applications with timed hierarchical state diagrams is indicated. By their translation to program code, schedulable objects are formed, representing the applications? semantics. A distinction is made between active and passive objects. Active objects communicate among each other and manage the passive objects and their co-operation. An active "super" object, called Configuration Manager, is foreseen for: the loading of the initial SW configuration onto its native processing node, any subsequent changes in the SW configuration, as well as inter-node co-operation.