Skip to Main Content
A new resiliency scheme for a distributed database system with replicated data is presented. One salient feature of the scheme is that it does not employ a log subsystem; hence, it can be used for a highly reliable system that must tolerate a total crash of a site. In addition, the scheme supports system partitioning without any consistency problem. The new scheme is based on a precise treatment of logical data; in particular, a precise definition of resilient system operation is given in terms of logical data.