Efficient Agreement Protocols for Asynchronous Distributed Systems

170 Pages·2017·5.74 MB·English

by  Izabela Moise

Efficient Agreement Protocols for Asynchronous
Distributed Systems
Izabela Moise
To cite this version:
Izabela Moise. Efficient Agreement Protocols for Asynchronous Distributed Systems. Distributed,
Parallel, and Cluster Computing [cs.DC]. Université Rennes 1, 2011. English. ￿NNT: ￿. ￿tel-
00658981v2￿
HAL Id: tel-00658981
https://theses.hal.science/tel-00658981v2
Submitted on 2 Feb 2012
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est
archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
No d’ordre: 4481 ANNÉE2011
THÈSE / UNIVERSITÉ DE RENNES 1
sous le sceau de l’Université Européenne de Bretagne
pour le grade de
DOCTEUR DE L’UNIVERSITÉ DE RENNES 1
Mention: INFORMATIQUE
Ecole doctorale MATISSE
présentée par
Izabela Moise
préparéeàl’unitéderechercheno 6074-IRISA
InstitutdeRechercheenInformatiqueetSystèmesAléatoires
IFSIC
Efficient
ThèsesoutenueàRennes
Agreement le12decembre2011
devantlejurycomposéde:
Protocols MichelHURFIN/directeurdethèse
Chargéderecherche,INRIARennes,France
in Asynchronous Jean-PierreLENARZUL/encadrant
Maitredeconférence,TélécomBretagneRennes,France
Distributed Systems RobertoBALDONI/rapporteur
Professeur,SapienzaUniversitàDiRoma,Rome,Italie
SébastienTIXEUIL/rapporteur
Professeur,LIP6,Paris,France
CésarVIHO/président
Professeur,UniversitédeRennes1,Rennes,France
AchourMOSTÉFAOUI/examinateur
Professeur,UniversitédeNantes,Nantes,France

“First you guess. Don’t laugh, this is the most important step. Then you compute the
consequences. Compare the consequences to experience. If it disagrees with experience,
the guess is wrong. In that simple statement is the key to science. It doesn’t matter how
beautiful your guess is or how smart you are or what your name is. If it disagrees with
experience,it’swrong. That’sallthereistoit.”
–RichardFeynman
“Ifweknewwhatitwasweweredoing,
itwouldnotbecalledresearch,wouldit?”
–AlbertEinstein

Acknowledgments
Thismanuscriptwouldnothaveseenthelightofdaywithoutmyadvisor,MichelHurfin.
I am most grateful for his support, patience and guidance throughout my PhD work. I am
veryfortunatetohavehadtheopportunitytoworkunderhissupervisionandmygratitude
goes beyond professional aspects. A great amount of thanks are for Jean-Pierre Le Narzul,
my second advisor. I was lucky enough to work with people from which I could learn a
lot and in a very pleasant manner, due to their great sense of humor. Thank you both for
everything.
Kind regards go as well to the members of the ADEPT team who made the working
environment a very pleasant one. Many thanks to my “second” team also, the members of
KerDatateam.
IamverygratefultoProf. CésarVihoforpresidingovermythesisdefenseandalsotothe
membersofthejury,Prof. RobertoBaldoni,Prof. SébastienTixeuil,Prof. AchourMostéfaoui
forevaluatingmyworkandprovidingusefulfeedback.
ManythankstoallmyfriendshereinRennes. Theirnamesaretoomanytobementioned
here. A simple “thank you” does not seem enough to express my gratitude for the help of
mygirls,mysisterDianaandAlexa. Thankyou,lesfilles.

v
Abstract
TheConsensusproblemisrecognizedasacentralparadigmoffault-tolerantdistributed
computing. In a purely asynchronous system, Consensus is impossible to solve in a de-
terministic manner. However, by enriching the system with some synchrony assumptions,
severalsolutionshavebeenproposedinordertocircumventtheimpossibilityresult,among
whichthePaxosapproachintroducedbyLamport.
Thisworkrepresentsacontributiontotheconstructionofefficientconsensusprotocolsin
asynchronousdistributedsystems. Thealgorithmiccontributionofthisthesisconsistsofan
efficientframework,calledthePaxos-MICprotocol,thatfollowsthePaxosapproachandin-
tegratestwoexistingoptimizations. Paxos-MICgeneratesasequenceofconsensusinstances
andguaranteesthepersistenceofalldecisionvalues. Themainfeatureoftheprotocolisits
adaptability. As one of the optimizations may be counterproductive, Paxos-MIC incorpo-
rates a triggering mechanism that dynamically enables the optimization. This mechanism
relies on several triggering criteria with the main purpose of predicting if the activation
of the optimization will be beneficial or not. Extensive experimentations on the Grid’5000
testbed were carried out with the purpose of evaluating the protocol and the efficiency of
thetriggeringcriteria.
A second part of this work focuses on the use of consensus as a building-block for de-
signing higher-level applications. We consider the particular context of transactional mo-
bileagentsandweproposeasolutiontosupporttheexecutionoftransactionsinanad-hoc
network. This solution relies on an everlasting sequence of decision values, generated by
repeatedly invoking a consensus building-block. The consensus service is provided by the
Paxos-MICframework.
vi
vii
Contents
1 Introduction 1
1.1 Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 RoadMap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Part I – Context: ConsensusinAsynchronousDistributedSystems 7
2 ConsensusinAsynchronousDistributedSystems 9
2.1 Systemmodel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Consensus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Theproblemspecification . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Agreementproblems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Impossibilityresult . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Unreliablefailuredetectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
♢
2.4 Sbasedprotocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4.1 Fundamentalprinciples . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4.2 TheChandra-Touegprotocol . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.3 TheEarlyConsensusprotocol. . . . . . . . . . . . . . . . . . . . . . . . 20
2.4.4 TheGeneralConsensusProtocol-AGeneralQuorum-BasedApproach 21
2.4.5 Particularities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5 Otherapproachestospecifyingthesystemmodel . . . . . . . . . . . . . . . . 23
2.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3 ThePaxosprotocol 25
3.1 LeaderElectionOracles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2 HistoryofPaxos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.1 TheoriginalPaxos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26