Carole Delporte-Gallet
ABSTRACT
In the synchronous round-based model, a process crash is dirty if it occurs exactly while a process is sending messages in a round, and this causes the process to send to some, but not all, of the intended recipients for the given round. Dirty crashes are possible; however, they are unlikely to occur, since the time spent sending messages is usually very small compared to the maximum message delay (i.e., compared to the duration of a round). In this paper, we investigate how fast one can solve some agreement problems, namely consensus and terminating reliable broadcast (TRB), when the number of dirty crashes that occur is small. In particular, we describe some algorithms for the uniform and non-uniform versions of these problems, and provide some matching lower bounds. All our uniform algorithms are strictly better than conventional early-stopping algorithms, in the sense that they never take more rounds to decide or halt, and they take fewer rounds when the number of dirty crashes is small.
- M. K. Aguilera and S. Toueg. A simple bivalency proof that t-resilient consensus requires t+1 rounds. Information Processing Letters, 71(3--4):155--158, August 1999.Google Scholar
Cross Ref
- P. Berman, J. A. Garay, and K. J. Perry. Optimal early stopping in distributed consensus (extended abstract). In WDAG'92: Proceedings of the 6th International Workshop on Distributed Algorithms, pages 221--237. Springer, 1992. Google ScholarDigital Library
- T. D. Chandra and S. Toueg. Time and message efficient reliable broadcasts. In WDAG '90: Proceedings of the 4th International Workshop on Distributed Algorithms, pages 289--303. Springer, 1990. Google ScholarDigital Library
- B. Charron-Bost and A. Schiper. Uniform consensus is harder than consensus. J. Algorithms, 51(1):15--37, 2004. Google ScholarDigital Library
- D. Dolev, R. Reischuk, and H. R. Strong. Early stopping in byzantine agreement. J. ACM, 37(4):720--741, 1990. Google ScholarDigital Library
- C. Dwork and Y. Moses. Knowledge and common knowledge in a byzantine environment: Crash failures. Information and Computation, 88(2):156--186, Oct 1990. Google ScholarDigital Library
- F. Le Fessant. The complexity of early-deciding in unreliable synchronous networks. Technical Report TR-2003-23, Microsoft Research, 2003.Google Scholar
- F. Fich and E. Ruppert. Hundreds of impossibility results for distributed computing. Distributed Computing, 16(2-3):121--163, 2003. Google ScholarDigital Library
- M. J. Fischer and N. A. Lynch. A lower bound for the time to assure interactive consistency. Information Processing Letters, 14(4):183--186, June 1982.Google ScholarCross Ref
- M. J. Fischer, N. A. Lynch, and M. S. Paterson. Impossibility of distributed consensus with one faulty process. J. ACM, 32(2):374--382, 1985. Google ScholarDigital Library
- V. Hadzilacos. A lower bound for byzantine agreement with fail-stop processors. Technical Report 21-83, Department of Computer Science, Harvard University, July 1983.Google Scholar
- V. Hadzilacos and S. Toueg. A modular approach to the specification and implementation of fault-tolerant broadcasts. Technical Report TR94-1425, Department of Computer Science, Cornell University, May 1994.Google Scholar
- I. Keidar and S. Rajsbaum. A simple proof of the uniform consensus synchronous lower bound. Information Processing Letters, 85(1):47--52, January 2003. Google ScholarDigital Library
- L. Lamport and M. Fischer. Byzantine generals and transaction commit protocols. Technical Report 62, SRI International and Yale University, April 1982.Google Scholar
- N. A. Lynch. Distributed Algorithms. Morgan Kaufmann Publishers Inc., 1996. Google ScholarDigital Library
- Y. Moses and S. Rajsbaum. A layered analysis of consensus. SIAM J. Comput., 31(4):989--1021, 2002. Google ScholarDigital Library
- A. Mostefaoui, S. Rajsbaum, and M. Raynal. Using conditions to expedite consensus in synchronous distributed systems. In DISC, volume 2848 of Lecture Notes in Computer Science, pages 249--263. Springer, 2003.Google Scholar
- Ph. Raïpin Parvédy and M. Raynal. Optimal early stopping uniform consensus in synchronous systems with process omission failures. In SPAA '04: Proceedings of the sixteenth annual ACM symposium on Parallelism in algorithms and architectures, pages 302--310. ACM Press, 2004. Google ScholarDigital Library
- M.-C. Roçu. Early-stopping terminating reliable broadcast protocol for general-omission failures. In PODC '96: Proceedings of the fifteenth annual ACM symposium on Principles of distributed computing, page 209. ACM Press, 1996. Google ScholarDigital Library
- J. Cao, X. B. Wang and Y. M. Teo. An approach to achieve message efficient early-stopping uniform consensus protocols. In ISPAN '04: Proceedings of International Symposium on Parallel Architectures, Algorithms, and Networks. IEEE Computer Society Press, May 2004.Google Scholar
Index Terms
- Fast fault-tolerant agreement algorithms
Recommendations
Uniform Agreement Despite Process Omission Failures
IPDPS '03: Proceedings of the 17th International Symposium on Parallel and Distributed ProcessingA process fails by omission if it "forgets" to send or receive messages. Considering omission failures is crucial for distributed systems, as such failures model both crash failures and incorrect behavior of process input/output buffers (such as buffer ...
Early-deciding consensus is expensive
PODC '13: Proceedings of the 2013 ACM symposium on Principles of distributed computingIn consensus, the n nodes of a distributed system seek to take a consistent decision on some output, despite up to t of them crashing or even failing maliciously, i.e., behaving "Byzantine''. It is known that it is impossible to guarantee that ...
Structured derivation of semi-synchronous algorithms
DISC'11: Proceedings of the 25th international conference on Distributed computingThe semi-synchronous model is an important middle ground between the synchronous and the asynchronous models of distributed computing. In this model, processes can detect (timeout) when other processes fail. However, since detection is done by timing ...
Comments