Resource sharing in a complex fault-tolerant system

Sharing of pooled hardware resources is examined as an efficient solution to the problem of ensuring fault tolerance in a complex real-time system. Efficiency is discussed in terms of both the number of hardware units required to implement a system and the amount of fault tolerance provided by the hardware. In the context of a general avionics application, the PRIMA architecture described applies resource sharing to processors, buses, and stores, Processors are pooled in groups preserving the topologies of multiprocessor architectures, while buses are pooled according to the processor groups which they serve. Stores are pooled by updating local memory to reflect all system updates. Local reconfiguration of the system within each processor class enables the system to detect and handle faults rapidly. Global reconfiguration of the system enables the system to maintain critical functionality despite multiple hardware failures