Database sharing (DB-sharing) refers to a loosely or closely ccupled multiprocessor architecture where all processors share a common database at the disk level. Such systems primarily aim at high availability and high performance demanded by large applications in online transaction processing. To achieve these goals a synchronization
technique is required that efficiently coordinates the processors' database accesses and that also works properly alter a processor crash. The described primary copy
algorithm seems to be a good candidate to meet these requirements since it permits flexible adaptation to changing working conditions. Besides of the basic protocol we
specify the recovery actions alter a processor crash. Mechanisms are proposed that allow the continuation of synchronization alter a crash with little interference to transaction processing.