D.D.E. Long and J.-F. Paris. Regeneration protocols for replicated objects. In Proceedings of the Fifth International Conference on Data Engineering, pages 538--545, Los Angeles, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....representative examples of each class to illustrate our classification scheme. Furthermore, we compare the structural differences of RC methods but do not analyze their performance due to the structural differences. This is in contrast to cost benefit analysis of particular RC algorithms such as [11, 32]. To the best of our knowledge, Ceri et al. s classification [12] is the only systematic attempt to analyze and compare the structure of RC algorithms. Specifically, they have analyzed RC protocols according to the number of copies read and written, in conjunction with the kind of agreement ....

....are statically allocated on certain sites. An alternative is to use some kind of copy location protocol (e.g. a directory or broadcast) to dynamically find the copies. This way, physical copies may be regenerated [39] when a site goes down, providing more availability than static RC protocols [32]. Note that regeneration of copies is independent from the issue of dynamic quorum formation in quorum consensus protocols. Regardless of how and when quorum is formed, copy regeneration is possible. Another possibility is to separate the replication of metadata from the replication of real data. ....

D.D.E. Long and J.-F. Paris. Regeneration protocols for replicated objects. In Proceedings of the Fifth International Conference on Data Engineering, pages 538--545, Los Angeles, 1989.

....in [LPC89] and the schemes considered are available copies, dynamic voting, linear dynamic voting and majority 8 consensus. It is shown that the available copies scheme has the highest reliability among the schemes studied. The availability of regeneration based protocols is considered in [LP89] The protocols studied are regeneration with m copies and n spares, regenerative available copies, and regenerative dynamic voting. The reliability of regeneration based protocols is analyzed in [LCS89] and the protocols studied are regenerative available copies, regenerative dynamic linear ....

Darrell D. E. Long and Jehan-Francois Paris. Regeneration protocols for replicated objects. In Proceedings of the 5th International Conference on Data Engineering, pages 538--545, 1989.

....at multiple proxy caches. Much of existing work on dynamic replication has concentrated on quorum based replica management. This work focuses on algorithms for adjusting quorums when replica sets change (e.g. 1, 21, 33] and on regenerating object replicas in place of failed ones (e.g. [27] ) In contrast, our work employs replication and migration for performance, and assumes that asynchronous update propagation and not quorums is used for consistency. Existing protocols for performance motivated dynamic replication rely on assumptions that are unrealistic in the Internet context. ....

D. D. E. Long and J.-F. Paris. Regeneration protocols for replicated objects. In Proc. of the 5th IEEE Int. Conf. on Data Eng., pp. 538-545, 1989.

....algorithms is to improve the availability of the DCS. A different approach to maintaining replicated databases has been adopted by proponents of a technique called regeneration, which is quite similar to file migration. Regeneration in distributed computing systems has been recently suggested in [19, 15, 14, 9, 1] as a mechanism for maintaining consistency of replicated data in distributed computing systems. Distributed databases and distributed file service mechanisms are examples of distributed computing systems. The algorithm proposed in [19] is an example of a regeneration algorithm that implements a ....

D. E. Long and J. L. Carroll. Regeneration protocols for replicated objects. In 5th IEEE Conference on Data Engineering, pages 538--545, 1989.

....algorithms is to improve the availability of the DCS. A different approach to maintaining replicated databases has been adopted by proponents of a technique called regeneration, which is quite similar to file migration. Regeneration in distributed computing systems has been recently suggested in [17, 13, 12, 7] as a mechanism for maintaining consistency of replicated data in distributed computing systems. Distributed databases and distributed file service mechanisms are examples of distributed computing systems. The algorithm proposed in [17] is an example of a regeneration algorithm that implements a ....

D. E. Long and J. L. Carroll. Regeneration protocols for replicated objects. In 5th IEEE Conference on Data Engineering, pages 538--545, 1989.

....WITNESSES The notion of regenerating replicas to replace those lost due to site failures was first introduced by the regeneration algorithm (RA) PNP88] which provided mutual and serial consistency of replicated objects in a partition free distributed system. Since then, it has been shown [LoPa89, LCS89] that regeneration is a generally applicable technique that can be profitably combined with many replication control protocols. In a partitionable distributed system, consensus protocols are required to maintain consistency. Regeneration has been combined with majority consensus voting to obtain ....

....system, consensus protocols are required to maintain consistency. Regeneration has been combined with majority consensus voting to obtain a simple protocol for maintaining mutual consistency, and has also been combined with dynamic voting protocols to provide an increased level of fault tolerance [LoPa89, LCS89]. The penalty for using regeneration with these protocols is an increase in both network traffic and peak storage requirements. A witness [Pari86] requires very little storage: only enough so that it can attest to the state of the replicated object. We describe a method that combines ....

D. D. E. Long and J.-F. Pa ris, "Regeneration Protocols for Replicated Objects," Proc. 5th International Conference on Data Engineering, (1989), pp. 538-545.

....WITNESSES The notion of regenerating replicas to replace those lost owing to site failures was first introduced by the regeneration algorithm (RA) 15] which provided mutual and serial consistency of replicated objects in a partition free distributed system. Since then, it has been shown [8, 10, 17] that regeneration is a generally applicable technique that can be profitably combined with many replication control protocols. In a partitionable distributed system, consensus protocols are required to maintain consistency. Regeneration has been combined with majority consensus voting to obtain a ....

....system, consensus protocols are required to maintain consistency. Regeneration has been combined with majority consensus voting to obtain a simple protocol for maintaining mutual consistency, and has also been combined with dynamic voting protocols to provide an increased level of fault tolerance [8, 10]. The penalty for using regeneration with these protocols is an increase in both network traffic and peak storage requirements. A witness [12] requires very little storage: only enough so that it can attest to the state of the replicated object. We describe a method that combines regeneration, a ....

D. D. E. Long and J.-F. Pa ris, "Regeneration Protocols for Replicated Objects," Proc. 5th Int. Conf. on Data Engineering, (1989), pp. 538-545.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC