Safety-critical systems have typically been designed in a static fashion, preventing run-time changes in the operational parameters that could possibly lead to unpredictable and unsafe operating scenarios. However, there is a growing demand for flexibility, mainly to support evolving requirements, simplify maintenance and repair, and improve the efficiency in using system resources. This last aspect is directly related with operational flexibility and impacts on the system cost, rising, thus, its interest. This paper discusses at a conceptual level the motivations and problems associated with introducing operational flexibility in distributed hard real-time systems with critical safety requirements. Then, it discusses the advantages arising from such flexibility and presents a general architecture that can be used to support it on such systems. This architecture has been the basis for a set of recent protocols, namely FTT-CAN and FTT-Ethernet.