System architecture for partition-tolerant distributed databases

An overview is presented of an approach to distributed database design which emphasizes high availability in the face of network partitions and other communication failures. This approach is appropriate for applications which require continued operation and can tolerate some loss of integrity of the data. Each site presents its users and application programs with the best possible view of the data which it can, based on those updates which it has received so far. Mutual consistency of replicated copies of data is ensured by using time stamps to establish a known total ordering on all updates issued, and by a mechanism which ensures the same final result regardless of the order in which a site actually receives these updates. A mechanism is proposed, based on alerters and triggers, by which applications can deal with exception conditions which may arise as a consequence of the high-availability architecture. A prototype system which demonstrates this approach is near completion.