Eventual Leader Service in Unreliable Asynchronous Systems: Why? How?

Raynal, Michel

doi:10.1109/nca.2007.19

Cited by 4 publications

(2 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Message-pattern approach: Given a query issued by p, the set of first RESP messages received by p to this query are denoted winning responses [6], [9]. A process p is P S−accessible (resp., ♦P S−accessible) if for τ 0 ≥ 0 (resp., ∃τ 0 , ∀τ ≥ τ 0 ) there exists a set Q of processes such that Q ∈ P S and p / ∈ Q(τ ) and a QU ERY message broadcast by p at τ receives RESP messages from processes of Q(τ ) and these responses are always (resp., eventually always) winning responses.…”

Section: Some Propertiesmentioning

confidence: 99%

Implementing a Flexible Failure Detector That Expresses the Confidence in the System

Rossetto

Geyer

Arantes

et al. 2016

2016 Seventh Latin-American Symposium on Dependable Computing (LADC)

View full text Add to dashboard Cite

Abstract-Traditional unreliable failure detectors are per process oracles that provide a list of nodes suspected of having failed. In [1], we introduced the Impact failure detector that outputs a trust level value which is the degree of confidence in the system. An impact factor is assigned to each node and the trust level is equal to the sum of the impact factors of the nodes not suspected to have failed. An input threshold parameter defines an impact factor limit value, over which the confidence degree on the system is ensured. The impact factor indicates the relative importance of the process in the set S, while the threshold offers a degree of flexibility for failures and false suspicions. We propose in this article two different algorithms, based on query-response message rounds, that implement the Impact FD whose conceptions were tailored to satisfy the Impact FD's flexibility. The first one exploits the time-free message pattern approach while the second one considers a set of bounded timely responses. We also introduced the concept that a process can be P S−accessible (or ♦P S−accessible) which guarantees that the system S will always (or eventually always) be trusted to this process as well as two properties, P R(IT ) and P R(♦IT ), that characterize the minimum necessary stability condition of S that ensures confidence (or eventual confidence) on it. In both implementations, if the process that monitors S is P S−accessible or ♦P S−accessible, at every query round, it only waits (or eventually only waits) for a set of response that satisfy the threshold. A crucial facet of this set of processes is that it is not fixed, i.e., the set of processes can change at each round, which is in accordance with the flexibility capacity of the Impact FD.

show abstract

Section: Some Propertiesmentioning

confidence: 99%

Implementing a Flexible Failure Detector That Expresses the Confidence in the System

Rossetto

Geyer

Arantes

et al. 2016

2016 Seventh Latin-American Symposium on Dependable Computing (LADC)

View full text Add to dashboard Cite

show abstract

“…This service is strictly weaker than the classical eventual leader service Ω [18], since we do not require that every correct site eventually outputs the same leader. An algorithm that returns to every process itself, trivially implements the Eventual Weak Leader Service.…”

Section: System Model and Assumptionsmentioning

confidence: 99%

Fault-Tolerant Partial Replication in Large-Scale Database Systems

Sutra

Shapiro

2008

Lecture Notes in Computer Science

View full text Add to dashboard Cite

We investigate a decentralised approach to committing transactions in a replicated database, under partial replication. Previous protocols either reexecute transactions entirely and/or compute a total order of transactions. In contrast, ours applies update values, and generate a partial order between mutually conflicting transactions only. It results that transactions execute faster, and distributed databases commit in small committees. Both effects contribute to preserve scalability as the number of databases and transactions increase. Our algorithm ensures serializability, and is live and safe in spite of faults.

show abstract