T.D. Chandra and S. Toueg. Unreliable failure detector for asynchronous distributed systems. In 10 th Annual ACM Symposium on Principles of Distributed Computing, pages 325--340, 1992.  Home/Search   Document Details and Download   Summary   ACM   TOC   Related Articles   Check

....We also brie#y discuss the MULTIPAXOS protocol and a data replication algorithm which uses MULTIPAXOS. With MULTIPAXOS the high availability of the replicated data is combined with high fault tolerance. 1.1. Related work The consensus algorithms of Dwork et al. 9] and of Chandra and Toueg [2] bear some similarities with PAXOS. The algorithm of [9] also uses rounds conducted by a leader, but the strategy used in each round is di#erent from the one used by PAXOS. Also, 9] does not consider process restarts. The time analysis provided in [9] is conditional on a global stabilization ....

....times and message delivery times satisfy the time assumptions. This is similar to our stabilized analysis. A similar time analysis, applied to the problem of reliable group communication, can be found in [12] The algorithm of Chandra and Toueg is based on the idea of an abstract failure detector [2]. It turns out that failure detectors provide an abstract and modular way of incorporating partial synchrony assumptions in the model of computation. A failure detector incorporates the partial synchrony considered in this paper. One of the algorithms in [2] uses a #S failure detector, which ....

[Article contains additional citation context not shown here]

T.D. Chandra, S. Toueg, Unreliable failure detector for asynchronous distributed systems, J. ACM 43 (2) (1996.

....problem cannot be solved even in the case where at most one process fails [18] If a reliable failure detector were provided then the consensus problem would be solvable. This clearly implies that in a completely asynchronous setting no reliable failure detector can be provided. Chandra and Toueg [5] gave a definition of unreliable failure detector, and characterized failure detectors in terms of two properties: completeness, which requires that the failure detector eventually suspect any stopped process, and accuracy, which restricts the mistakes a failure detector can make. No failure ....

....failure detector, and characterized failure detectors in terms of two properties: completeness, which requires that the failure detector eventually suspect any stopped process, and accuracy, which restricts the mistakes a failure detector can make. No failure detectors are actually implemented in [5]. The failure detector provided in this thesis differs from those classified by Chandra and Toueg in that it provides reliability conditional to the system stabilization. If the system eventually stabilizes then our failure detector can be classified in the class of the eventually perfect failure ....

T.D. Chandra, S. Toueg, Unreliable failure detector for asynchronous distributed systems, Proceedings of PODC 91, pp. 325--340. in Proceedings of the 10 th Annual ACM Symposium on Principles of Distributed Computing, pp. 325--340, August 1991.

....scenario. We also discuss the multipaxos protocol and provide a data replication algorithm using multipaxos. With multipaxos the high availability of the replicated data is combined with high fault tolerance. Related work. The consensus algorithms of Dwork et al. 5] and of Chandra and Toueg [2] bear some similarities with paxos. The algorithm of [5] also uses rounds conducted by a leader, but the rounds are conducted sequentially, whereas in paxos a leader can start a round at anytime and multiple leaders are allowed. The strategy used in each round by the algorithm of [5] is different ....

....communication, can be found in [6] The algorithm of Chandra and Toueg is based on the idea of an abstract failure detector. It turns out that failure detectors provide an abstract and modular way of incorporating partial synchrony assumptions in the model of computation. One of the algorithms in [2] uses the failure detector PiS which incorporates the partial synchrony considered in this paper. That algorithm is based on the rotating coordinator paradigm and as paxos uses majorities to achieve consistency. The performances of the Toueg and Chandra algorithm and of the paxos algorithm seem ....

T.D. Chandra, S. Toueg, Unreliable failure detector for asynchronous distributed systems, Journal of the ACM, Vol. 43 (2), pp. 225--267. A preliminary version appeared in the Proceedings of the 10 th Annual ACM Symposium on Principles of Distributed Computing, pages 325--340, August 1991.

No context found.

T.D. Chandra and S. Toueg. Unreliable failure detector for asynchronous distributed systems. In 10 th Annual ACM Symposium on Principles of Distributed Computing, pages 325--340, 1992.

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