How to Choose a Timing Model

When employing a consensus algorithm for state machine replication, should one optimize for the case that all communication links are usually timely or for fewer timely links? Does optimizing a protocol for better message complexity hamper the time complexity? In this paper, we investigate these types of questions using mathematical analysis as well as experiments over PlanetLab (WAN) and a LAN. We present a new and efficient leader-based consensus protocol that has O(n) stable-state message complexity (in a system with n processes) and requires only O(n) links to be timely at stable times. We compare this protocol with several previously suggested protocols. Our results show that a protocol that requires fewer timely links can achieve better performance, even if it sends fewer messages.

[1]  David E. Culler,et al.  Operating Systems Support for Planetary-Scale Network Services , 2004, NSDI.

[2]  Mikel Larrea,et al.  Optimal implementation of the weakest failure detector for solving consensus , 2000, Proceedings 19th IEEE Symposium on Reliable Distributed Systems SRDS-2000.

[3]  Michel Raynal,et al.  Eventual Leader Election with Weak Assumptions on Initial Knowledge, Communication Reliability, and Synchrony , 2006, DSN.

[4]  Ulrich Schmid.  How to model link failures: a perception-based fault model , 2001, 2001 International Conference on Dependable Systems and Networks.

[5]  Danny Dolev,et al.  Evaluating Total Order Algorithms in WAN , 2003 .

[6]  Idit Keidar,et al.  The Overhead of Indulgent Failure Recovery , 2006 .

[7]  Rachid Guerraoui,et al.  Fast Indulgent Consensus with Zero Degradation , 2002, EDCC.

[8]  Nancy A. Lynch,et al.  An introduction to input/output automata , 1989 .

[9]  Rachid Guerraoui.  Revistiting the Relationship Between Non-Blocking Atomic Commitment and Consensus , 1995, WDAG.

[10]  Sam Toueg,et al.  The weakest failure detector for solving consensus , 1996, JACM.

[11]  Idit Keidar,et al.  Timeliness, failure-detectors, and consensus performance , 2006, PODC '06.

[12]  Nancy A. Lynch,et al.  Consensus in the presence of partial synchrony , 1988, JACM.

[13]  Péter Urbán,et al.  Comparison of failure detectors and group membership: performance study of two atomic broadcast algorithms , 2003, 2003 International Conference on Dependable Systems and Networks, 2003. Proceedings..

[14]  Ulrich Schmid.  Failure Model Coverage under Transient Link Failures , 2008 .

[15]  Dahlia Malkhi,et al.  Omega Meets Paxos: Leader Election and Stability Without Eventual Timely Links , 2005, DISC.

[16]  Leslie Lamport,et al.  The Implementation of Reliable Distributed Multiprocess Systems , 1978, Comput. Networks.

[17]  Rachid Guerraoui,et al.  The information structure of indulgent consensus , 2004, IEEE Transactions on Computers.

[18]  Leslie Lamport,et al.  The part-time parliament , 1998, TOCS.

[19]  Rachid Guerraoui,et al.  The overhead of consensus failure recovery , 2007, Distributed Computing.

[20]  Marcos K. Aguilera,et al.  Stable Leader Election , 2001, DISC.

[21]  Rachid Guerraoui,et al.  How fast can eventual synchrony lead to consensus? , 2005, 2005 International Conference on Dependable Systems and Networks (DSN'05).

[22]  Idit Keidar,et al.  Evaluating the running time of a communication round over the internet , 2002, PODC '02.

[23]  H. Chernoff.  A Measure of Asymptotic Efficiency for Tests of a Hypothesis Based on the sum of Observations , 1952 .

[24]  Flaviu Cristian,et al.  The Timed Asynchronous Distributed System Model , 1999, IEEE Trans. Parallel Distributed Syst..