Divergence control; Freshness control;Incoherency bounds
In a distributed system, information is often replicated with copies of the same data stored on several sites. Ideally, all copies would be kept identical, but doing this imposes a performance penalty. Many system designs allow replicas to lag behind the latest value. For some applications, it is acceptable to use out-of-date copies, provided they are not too far from the true, current value. Freshness refers to a measure of the difference between a replica and the current value.
The tradeoff between consistency and performance or availability is an old theme in distributed computing. In the database community, many researchers worked on ideas connected with explicitly allowing some discrepancy between replicas during the late 1980s and early 1990s. Early papers identified many of the diverse freshness measures discussed here, from groups at Princeton, Bellcore and Stanford [1, 10, 11]....
- 2.Bernstein PA, Fekete A, Guo H, Ramakrishnan R, Tamma P. Relaxed-currency serializability for middle-tier caching and replication. In: Proceedings of the ACM SIGMOD International Conference on Management of Data; 2006. p. 599–610.Google Scholar
- 3.Guo H, Larson PÅ, Ramakrishnan R, Goldstein J. Relaxed currency and consistency: how to say “good enough” in SQL. In: Proceedings of the ACM SIGMOD International Conference on Management of Data; 2004. p. 815–26.Google Scholar
- 10.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
- 11.Wiederhold G, Qian X. Consistency control of replicated data in federated databases. In: Proceedings of the Workshop on the management of replicated data. Houston. 1990. p. 130–2.Google Scholar