Approaches to implementation of a repairable distributed recovery block scheme

The authors previously proposed (1984) the basic concept of the distributed recovery block (DRB) scheme as an approach to uniform treatment of hardware and software faults in real-time applications. Design issues that arise in implementing the DRB scheme are discussed together with some promising approaches. Issues in extending the DRB scheme with the capability of reincorporating a repaired node without disrupting the real-time computing service are also discussed. An experimental implementation of the repairable DRB scheme into a real-time distributed computer system (DCS) testbed and subsequent measurement of the system performance demonstrated the fast forward recovery capability and the logical soundness of the scheme.<<ETX>>