Enhanced fault-tolerance through Byzantine failure detection

Rida Bazzi, Maurice Herlihy

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

We consider a variant of the Byzantine failure model in which Byzantine processes are eventually detected and silenced, and investigate the fault-tolerance of the classical broadcast and agreement problems. We show that if all Byzantine processes are eventually detected, then it is possible to solve the broadcast problem in the presence of any number of Byzantine processes. If only a fraction of the Byzantine processes can be detected, then we show that it is possible to solve consensus (and broadcast) if the total number of processes is N ≥ 2f + 3F + 1, where f is the number of Byzantine processes that are eventually detected and F is the number of those that are never detected. We show that 2f + 3F + 1 is a lower bound to solve the consensus and broadcast problems.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages129-143
Number of pages15
Volume5923 LNCS
DOIs
StatePublished - 2009
Event13th International Conference on Principles of Distributed Systems, OPODIS 2009 - Nimes, France
Duration: Dec 15 2009Dec 18 2009

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5923 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other13th International Conference on Principles of Distributed Systems, OPODIS 2009
CountryFrance
CityNimes
Period12/15/0912/18/09

Fingerprint

Failure Detection
Fault tolerance
Fault Tolerance
Broadcast
Lower bound

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Bazzi, R., & Herlihy, M. (2009). Enhanced fault-tolerance through Byzantine failure detection. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 5923 LNCS, pp. 129-143). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5923 LNCS). https://doi.org/10.1007/978-3-642-10877-8_12

Enhanced fault-tolerance through Byzantine failure detection. / Bazzi, Rida; Herlihy, Maurice.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5923 LNCS 2009. p. 129-143 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5923 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Bazzi, R & Herlihy, M 2009, Enhanced fault-tolerance through Byzantine failure detection. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 5923 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 5923 LNCS, pp. 129-143, 13th International Conference on Principles of Distributed Systems, OPODIS 2009, Nimes, France, 12/15/09. https://doi.org/10.1007/978-3-642-10877-8_12
Bazzi R, Herlihy M. Enhanced fault-tolerance through Byzantine failure detection. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5923 LNCS. 2009. p. 129-143. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-10877-8_12
Bazzi, Rida ; Herlihy, Maurice. / Enhanced fault-tolerance through Byzantine failure detection. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5923 LNCS 2009. pp. 129-143 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{3941cc71570e409181e757fc1e293605,
title = "Enhanced fault-tolerance through Byzantine failure detection",
abstract = "We consider a variant of the Byzantine failure model in which Byzantine processes are eventually detected and silenced, and investigate the fault-tolerance of the classical broadcast and agreement problems. We show that if all Byzantine processes are eventually detected, then it is possible to solve the broadcast problem in the presence of any number of Byzantine processes. If only a fraction of the Byzantine processes can be detected, then we show that it is possible to solve consensus (and broadcast) if the total number of processes is N ≥ 2f + 3F + 1, where f is the number of Byzantine processes that are eventually detected and F is the number of those that are never detected. We show that 2f + 3F + 1 is a lower bound to solve the consensus and broadcast problems.",
author = "Rida Bazzi and Maurice Herlihy",
year = "2009",
doi = "10.1007/978-3-642-10877-8_12",
language = "English (US)",
isbn = "3642108768",
volume = "5923 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "129--143",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Enhanced fault-tolerance through Byzantine failure detection

AU - Bazzi, Rida

AU - Herlihy, Maurice

PY - 2009

Y1 - 2009

N2 - We consider a variant of the Byzantine failure model in which Byzantine processes are eventually detected and silenced, and investigate the fault-tolerance of the classical broadcast and agreement problems. We show that if all Byzantine processes are eventually detected, then it is possible to solve the broadcast problem in the presence of any number of Byzantine processes. If only a fraction of the Byzantine processes can be detected, then we show that it is possible to solve consensus (and broadcast) if the total number of processes is N ≥ 2f + 3F + 1, where f is the number of Byzantine processes that are eventually detected and F is the number of those that are never detected. We show that 2f + 3F + 1 is a lower bound to solve the consensus and broadcast problems.

AB - We consider a variant of the Byzantine failure model in which Byzantine processes are eventually detected and silenced, and investigate the fault-tolerance of the classical broadcast and agreement problems. We show that if all Byzantine processes are eventually detected, then it is possible to solve the broadcast problem in the presence of any number of Byzantine processes. If only a fraction of the Byzantine processes can be detected, then we show that it is possible to solve consensus (and broadcast) if the total number of processes is N ≥ 2f + 3F + 1, where f is the number of Byzantine processes that are eventually detected and F is the number of those that are never detected. We show that 2f + 3F + 1 is a lower bound to solve the consensus and broadcast problems.

UR - http://www.scopus.com/inward/record.url?scp=72749110089&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=72749110089&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-10877-8_12

DO - 10.1007/978-3-642-10877-8_12

M3 - Conference contribution

SN - 3642108768

SN - 9783642108761

VL - 5923 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 129

EP - 143

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -