Consistency Models for Replicated Data
Memory consistency; Replica consistency
When a distributed database system keeps several copies or replicas for a data item, at different sites, then the system may ensure that the copies are always consistent (that is, they have the same value), or the system may allow temporary discrepancy between the copies. Even if the copies are not the same, the algorithms that manage the data may be able to hide the discrepancies from clients. A consistency model defines the extent to which discrepancies can exist or be observed, between the copies. If the system offers a strong consistency model, then clients will not be aware of the fact that the system has replicated data, while a weak consistency model requires more careful programming of the clients, so they can cope with the discrepancies they observe.
Most replication research in the 1970s aimed to provide the illusion of an unreplicated database offering serializability. In the early 1980s,...
- 3.Demers AJ, Greene DH, Hauser C, Irish W, Larson J, Shenker S, Sturgis HE, Swinehart DC, Terry DB. Epidemic algorithms for replicated database maintenance. In: Proceedings of the ACM SIGACT-SIGOPS 6th Symposium on the Principles of Distributed Computing; 1987. p. 1–12.Google Scholar
- 4.Plattner C, Ganymed GA., Scalable replication for transactional web applications. In: Proceedings of the ACM/IFIP/USENIX 5th International Middleware Conference; 2004. p. 155–74.Google Scholar
- 5.Sheth AP, Rusinkiewicz M. Management of interdependent data: specifying dependency and consistency requirements. In: Proceedings of the Workshop on the Management of Replicated Data; 1990. p. 133–6.Google Scholar