| Pari86 J.-F. Pa ris, "Voting with Witnesses: A Consistency Scheme for Replicated Files," Proc. 6th Int. Conf. on Distributed Computing Systems (1986), pp. 606-612. |
....in replica availability and network topology. Both greatly improve availability when applied to replicated objects composed of more than three replicas, but do not perform significantly better 2 than majority consensus voting when fewer than four replicas are present. Voting with witnesses [Pari86, BeCi89] reduces the number of full replicas necessary to achieve a given level of fault tolerance by replacing some replicas by witnesses that hold no data but can attest to the status of the replicated object. Agrawal and El Abbadi [AgEl88] have recently proposed storing overlapping fragments of the ....
....network traffic and incurs no significant storage cost. Finally, regenerable volatile witnesses can be combined with two tier voting to provide a high level of fault tolerance with as few as two replicas in stable storage. Unlike other replication control protocols that require only two replicas [BeGo84, AgEl88, Pari86, ReTa88, Pari89], our novel technique requires only two sites providing stable storage, operates correctly in the presence of communication failures, and does not require any knowledge of the network topology. The remainder of this paper is organized as follows: Section two introduces regenerable volatile ....
[Article contains additional citation context not shown here]
J.-F. Pa ris, "Voting with Witnesses: A Consistency Scheme for Replicated Files," Proc. 6th International Conference on Distributed Computing Systems, (1986), pp. 606-612.
....in replica availability and network topology. Both protocols greatly improve the availability and reliability of replicated objects with more than three replicas. They do not perform better than Majority Consensus Voting (MCV) when only two or three replicas are present. Voting with Witnesses [Pari86a, Pari86b, BMP87] reduces the number of physical copies necessary to achieve a given level of fault tolerance by replacing some replicas by records of the object status that hold no data. Storage requirements can be significantly reduced but three voting entities are still required to improve upon the availability ....
....message traffic as most sites holding replicas can be expected to be operational most of the time. 3. AVAILABILITY ANALYSIS In this section we present an analysis of the availability provided by our protocol. Several definitions of the availability of a replicated object have been proposed [Pari86a, JaMu88]. We will assume here that the availability of a replicated data object is the stationary probability of the object being in a state permitting access. A S o (n) will denote the availability for an operation o of an object with n replicas managed by the protocol S. Our model consists of a set ....
[Article contains additional citation context not shown here]
J.-F. Pa ris, "Voting with Witnesses: A Consistency Scheme for Replicated Files," Proc. 6th International Conference on Distributed Computing Systems, (1986), pp. 606-612.
....changes in replica availability and network topology. Both greatly improve availability when applied to replicated objects composed of more than three replicas, but do not perform significantly better than majority consensus voting when fewer than four replicas are present. Voting with witnesses [4, 12] reduces the number of full replicas necessary to achieve a given level of faulttolerance by replacing some replicas by witnesses that hold no data but can attest to the status of the replicated object. Agrawal and El Abbadi [2] have recently proposed storing overlapping fragments of the object ....
....network traffic and incurs no significant storage cost. Finally, regenerable volatile witnesses can be combined with two tier voting to provide a high level of fault tolerance with as few as two replicas in stable storage. Unlike other replication control protocols that require only two replicas [2, 3, 12, 13, 16], our novel technique requires only two sites providing stable storage, operates correctly in the presence of communication failures, and does not require any knowledge of the network topology. The remainder of this paper is organized as follows: Section two introduces regenerable volatile ....
[Article contains additional citation context not shown here]
J.-F. Pa ris, "Voting with Witnesses: A Consistency Scheme for Replicated Files," Proc. 6th Int. Conf. on Distributed Computing Systems, (1986), pp. 606612.
....we present an analysis of the availability provided by our protocol and compare it to that afforded by the dynamic linear voting,which is the best pessimistic protocol that does not regenerate witnesses or replicas. Several definitions of the availability of a replicated data have been proposed [6, 9]. We will assume here that the availability of a replicated file is the stationary probability of the file being in a state permitting access. Our model consists of a set of nodes with independent failure modes that are connected via a network composed of network segments linked by gateways or ....
....Diagram for a Node and its Gateway 1 0 l=2l = 1 l 2 (b) State Transition Diagram for the Equivalent Aggregate Site Figure5:Aggregating a Node with its Gateway state analysis to be tractable. They have been made in most recent probabilistic analyses of the availability of replicated data [5, 7, 9]. Purely combinational models that do not require assumptions about failure and repair distributions have been proposed [12 13] but these models cannot distinguish between live and comatose replicas. We will consider six possible replica configurations. They were selected to provide a good sample ....
J.-F.Pa ris, "Voting with Witnesses: A Consistency Scheme for Replicated Files," Proc. 6th ICDCS,(1986), pp. 606-612.
....and repair followed Poisson processes, that is, that the distributions for TTF and TTR were exponential. This assumption is often made more for analytic simplicity than out of a conviction that it is the best model of reality. For example, some analyses of replication protocols using Markov models [P ris86, Long89] depend on that assumption. We investigate the accuracy of this assumption in 3.1.1 and 3.2.1. Availability is the fraction of time a system is functioning. More precisely, it is the stationary probability of the system being in a state where it can be accessed. For a replication protocol, for ....
J.-F. Pâris. Voting with witnesses: a consistency scheme for replicated files. In Proceedings of the 6th International Conference on Distributed Computing Systems, pages 606--612, Cambridge, 1986.
....and repair followed Poisson processes, that is, that the distributions for TTF and TTR were exponential. This assumption is often made more for analytic simplicity than out of a conviction that it is the best model of reality. For example, some analyses of replication protocols using Markov models [P ris86, Long89] depend on that assumption. We investigate the accuracy of this assumption in 3.1.1 and 3.2.1. Availability is the fraction of time a system is functioning. More precisely, it is the stationary probability of the system being in a state where it can be accessed. For a replication protocol, for ....
J.-F. Pâris. Voting with witnesses: a consistency scheme for replicated files. In Proceedings of the 6th International Conference on Distributed Computing Systems, pages 606--612, Cambridge, 1986.
....protocols greatly improve the availability and reliability of replicated data with more than two replicas. The generalized quorum consensus protocol (GQC) 9] exploits available type specific properties of data files to allow more flexible write quorum assignments. Vo ting with witnesses (VWW) [12], voting with ghosts (VWG) 17] and voting with bystanders (VWB) 13] share the common thread of introducing auxiliary entities that are used by the protocol to improve the availability of the replicated data. While these approaches significantly ease quorum requirements, they still provide lower ....
J.-F.Pa ris, "Voting with Witnesses: A Consistency Scheme for Replicated Files," Proc. 6th ICDCS (1986), pp. 606-612.
No context found.
Pari86 J.-F. Pa ris, "Voting with Witnesses: A Consistency Scheme for Replicated Files," Proc. 6th Int. Conf. on Distributed Computing Systems (1986), pp. 606-612.
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