In redundant real-time systems, the time required to exchange data and restore computation after a fault can affect availability and data integrity. Inter-computer DMA, checkpointing and hardware-assisted reconfiguration offer a coordinated approach to recovery.