Thông tin tài liệu
Lecture Notes in Computer Science 5923
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David Hutchison
Lancaster University, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Alfred Kobsa
University of California, Irvine, CA, USA
Friedemann Mattern
ETH Zurich, Switzerland
John C. Mitchell
Stanford University, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
Oscar Nierstrasz
University of Bern, Switzerland
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
TU Dortmund University, Germany
Madhu Sudan
Microsoft Research, Cambridge, MA, USA
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max-Planck Institute of Computer Science, Saarbruecken, Germany
Tarek Abdelzaher Michel Raynal
Nicola Santoro (Eds.)
Principles of
Distributed Systems
13th International Conference, OPODIS 2009
Nîmes, France, December 15-18, 2009
Proceedings
13
Volume Editors
Tarek Abdelzaher
University of Illinois at Urbana Champaign
Department of Computer Science
Urbana, IL 61801, USA
E-mail: zaher@cs.uiuc.edu
Michel Raynal
Université de Rennes1
IRISA
Campus de Beaulieu
Avenue du Général Leclerc
35042 Rennes Cedex, France
E-mail: raynal@irisa.fr
Nicola Santoro
Carleton University
School of Computer Science
1125 Colonel By Drive
Ottawa K1S 5B6, Canada
E-mail: santoro@scs.carleton.ca
Library of Congress Control Number: 2009939927
CR Subject Classification (1998): C.2.4, C.1.4, C.2.1, D.1.3, D.4.2, E.1, H.2.4
LNCS Sublibrary: SL 1 – Theoretical Computer Science and General Issues
ISSN
0302-9743
ISBN-10
3-642-10876-8 Springer Berlin Heidelberg New York
ISBN-13
978-3-642-10876-1 Springer Berlin Heidelberg New York
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting,
reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965,
in its current version, and permission for use must always be obtained from Springer. Violations are liable
to prosecution under the German Copyright Law.
springer.com
© Springer-Verlag Berlin Heidelberg 2009
Printed in Germany
Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India
Printed on acid-free paper SPIN: 12808168 06/3180 543210
Preface
OPODIS, the International Conference on Principles of Distributed Systems, is
an annual forum for presentation of state-of-the-art knowledge on principles of
distributed computing systems, including theory, design, analysis, implementa-
tion and application of distributed systems, among researchers from around the
world. The 13th edition of OPODIS was held during December 15–18, in Nimes,
France.
There were 71 submissions, and this volume contains the 23 regular contri-
butions and the 4 brief annoucements selected by the Progam Committee. All
submitted papers were read and evaluated by three to five PC members assisted
by external reviewers. The final decision regarding every paper was taken after
long discussions through EasyChair.
This year the Best Paper Award was shared by two papers: “On the Com-
putational Power of Shared Objects” by Gadi Taubenfeld and “Transactional
Scheduling for Read-Dominated Workloads” by Hagit Attiya and Alessia Milani.
The Best Student Paper Award was given to the paper “Decentralized Polling
with Respectable Participants” co-authored Kevin Huguenin and Maxime Monod
and their advisors.
The conference also featured two very interesting invited talks by Anne-Marie
Kermarrec and Maurice Herlihy. Anne-Marie’s talk was on “Navigating Web 2.0
with Gossple” and Maurice’s talk was on “Transactional Memory Today: A
Status Report.”
OPODIS has now found its place among the international conferences related
to principles of distributed computing and distributed systems. We hope that this
13th edition will contribute to the growth and the development of the conference
and continue to increase its visibility.
Finally we would like to thank Nicola Santoro, Conference General Chair,
Hac`ene Fouchal, Steering Committee Chair, and Bernard Thibault for their con-
stant help.
October 2009 Tarek Abdelzaher
Michel Raynal
Organization
General Chair
Nicola Santoro Carleton University, Canada
Program Committee Co-chairs
Tarek Abdelzaher University of Illinois at Urbana Champaign,
USA
Michel Raynal IRISA Rennes, France
Program Committee
Tarek Abdelzaher University of Illinois at Urbana Champaign,
USA (Co-chair)
Marcos Aguilera Microsoft, USA
James Anderson University of North-Carolina, USA
Jean Arlat LAAS, Toulouse, France
Hagit Attiya Technion, Israel
Theodore P. Baker Florida State University, USA
Roberto Baldoni University of Roma1, Italy
Gregor v. Bochmann University of Ottawa, Canada
Wei-ge Chen Microsoft, Beijing, China
UmaMaheswari Devi IBM Research Laboratory, India
Stefan Dobrev Slovak Academy of Science, Slovakia
Antonio Fern´andez University Rey Juan Carlos, Spain
Christof Fetzer Dresden University, Germany
Vijay K. Garg University of Texas at Austin/IBM, USA
Cyril Gavoille University of Bordeaux, France
M. Gonzalez Harbour University of Cantabria, Spain
Joel Goossens U.L.B, Belgium
Fabiola Greve U.F. Bahia, Brazil
Rachid Guerraoui EPFL, Switzerland
Herv´e Guyennet University of Franche-Comt´e, France
Ralf Klasing CNRS, Bordeaux, France
Xenofon Koutsoukos Venderbilt University, USA
Danny Krizanc Wesleyan University, USA
Chenyang Lu Washington University, USA
Marina Papatriantafilou Chalmers University of Technology, Sweden
Andrzej Pelc University of Quebec, Canada
Michel Raynal IRISA Rennes, France (Co-chair)
VIII Organization
Binoy Ravindran Virginia Tech, USA
Luis Rodrigues INESC-ID/IST, Portugal
Pierre Sens University Pierre et Marie Curie, France
Paul Spirakis Patras University, Greece
Gadi Taubenfeld Interdisiplinary Center, Israel
Eduardo Tovar ISEP-IPP, Portugal
Sebastien Tixeuil University Pierre et Marie Curie, France
Maarten Van Steen Amsterdam University, The Netherlands
Marko Vukolic IBM, Zurich, Switzerland
Kamin Whitehouse University of Vivgirid, USA
Masafumi Yamashita Kyushu University, Japan
Web and Publicity Chair
Thibault Bernard University of Reims Champagne-Ardenne,
France
Organizing Committee
Martine Couderc University of Nˆımes, France
Alain Findeli University of Nˆımes, France
Mostafa Hatimi University of Nˆımes, France
Dominique Lassarre University of Nˆımes, France
Thiery Spriet University of Avignon, France
Steering Committee
Tarek Abdelzaher University of Illinois at Urbana Champaign,
USA
Alain Bui University of Versailles St. Q. en Y., France
Marc Bui EPHE, France
Hacene Fouchal University of Antilles-Guyane, France (Chair)
Roberto Gomez ITESM-CEM, Mexico
Michel Raynal IRISA Rennes, France
Nicola Santoro Carleton University, Canada
Sebastien Tixeuil University of Pierre et Marie Curie, France
Philippas Tsigas Chalmers University of Technology, Sweden
External Referees
Isaac Amundson
Bjorn Andersson
Luciana Arantes
Shah Asaduzzaman
Roberto Beraldi
Jaiganesh Balasubramanian
Bharath Balasubramanian
Diogo Becker
Xiaohui Bei
Bjoern Brandenburg
Andrey Brito
Yann Busnel
Organization IX
Daniel Cederman
Ioannis Chatzigiannakis
Octav Chipara
Stephan Creutz
Shantanu Das
Jyotirmoy Deshmukh
UmaMaheswari Devi
Jos Mara Drake
Lcia Drummond
Philippe Duchon
Aida Ehyaei
Glenn Elliott
Robert Elsasser
Emeka Eyisi
Luis Lino Ferreira
Chien-Liang Fok
Hossein Fotouhi
Leszek Gasieniec
Gilles Geeraerts
Giorgos Georgiadis
Sascha Grau
JosCarlosPalenciaGutirrez
Greg Hackmann
Kai Han
David Hay
Phuong Ha Hoai
Michel Hurfin
Bijoy Jose
Manish Kushwaha
Shouwen Lai
Heath LeBlanc
Joao Leitao
Hennadiy Leontyev
Giorgia Lodi
Adnan Mian
Othon Michail
Alessia Milani
Neeraj Mittal
Jose Mocito
Alfredo Navarra
Nicolas Nisse
Martin Nowack
Vinit Ogale
Stephen Olivier
Filipe Pacheco
Guanhong Pei
Lucia Draque Penso
Shashi Prabh
Guido Proietti
Ying Qiao
Leonardo Querzoni
Tomasz Radzik
Carlos Ribeiro
Torvald Riegel
Mario Aldea Rivas
Mariusz Rokicki
Paulo Romano
Kunihiko Sadakane
Abusayeed Saifullah
Roopsha Samanta
Andre Schmitt
Christopher Thraves
Corentin Travers
Maryam Vahabi
Stefan Weigert
Jialin Zhang
Bo Zhang
Yuanfang Zhang
Dakai Zhu
Table of Contents
Invited Talks
Transactional Memory Today: A Status Report 1
Maurice Herlihy
Navigating the Web 2.0 with Gossple 2
Anne-Marie Kermarrec
Distributed Scheduling
Transactional Scheduling for Read-Dominated Workloads 3
HagitAttiyaandAlessiaMilani
Performance Evaluation of Work Stealing for Streaming Applications 18
Jonatha Anselmi and Bruno Gaujal
Not All Fair Probabilistic Schedulers Are Equivalent 33
Ioannis Chatzigiannakis, Shlomi Dolev, S´andor P. Fekete,
Othon Michail, and Paul G. Spirakis
Brief Announcement: Relay: A Cache-Coherence Protocol for
Distributed Transactional Memory 48
Bo Zhang and Binoy Ravindran
Distributed Robotics
Byzantine Convergence in Robot Networks: The Price of Asynchrony 54
Zohir Bouzid, Maria Gradinariu Potop-Butucaru, and
S´ebastien Tixeuil
Deaf, Dumb, and Chatting Asynchronous Robots: Enabling Distributed
Computation and Fault-Tolerance among Stigmergic Robots 71
Yoann Dieudonn´e, Shlomi Dolev, Franck Petit, and Michael Segal
Synchronization Helps Robots to Detect Black Holes in Directed
Graphs 86
Adrian Kosowski, Alfredo Navarra, and Cristina M. Pinotti
Fault and Failure Detection
The Fault Detection Problem 99
Andreas Haeberlen and Petr Kuznetsov
XII Table of Contents
The Minimum Information about Failures for Solving Non-local Tasks
in Message-Passing Systems 115
Carole Delporte-Gallet, Hugues Fauconnier, and Sam Toueg
Enhanced Fault-Tolerance through Byzantine Failure Detection 129
Rida A. Bazzi and Maurice Herlihy
Wireless and Social Networks
Decentralized Polling with Respectable Participants 144
Rachid Guerraoui, K´evin Huguenin, Anne-Marie Kermarrec, and
Maxime Monod
Efficient Power Utilization in Multi-radio Wireless Ad Hoc Networks 159
Roy Friedman and Alex Kogan
Adversarial Multiple Access Channel with Individual Injection Rates 174
Lakshmi Anantharamu, Bogdan S. Chlebus, and Mariusz A. Rokicki
Synchronization
NB-FEB: A Universal Scalable Easy-to-Use Synchronization Primitive
for Manycore Architectures 189
Phuong Hoai Ha, Philippas Tsigas, and Otto J. Anshus
Gradient Clock Synchronization Using Reference Broadcasts 204
Fabian Kuhn and Rotem Oshman
Brief Announcement: Communication-Efficient Self-stabilizing
Protocols for Spanning-Tree Construction 219
Toshimitsu Masuzawa, Taisuke Izumi, Yoshiaki Katayama, and
Koichi Wada
Storage Systems
On the Impact of Serializing Contention Management on STM
Performance 225
Tomer Heber, Danny Hendler, and Adi Suissa
On the Efficiency of Atomic Multi-reader, Multi-writer Distributed
Memory 240
Burkhard Englert, Chryssis Georgiou, Peter M. Musial,
Nicolas Nicolaou, and Alexander A. Shvartsman
Abortable Fork-Linearizable Storage 255
Matthias Majuntke, Dan Dobre, Marco Serafini, and Neeraj Suri
Table of Contents XIII
Distributed Agreement
On the Computational Power of Shared Objects 270
Gadi Taubenfeld
Weak Synchrony Models and Failure Detectors for Message Passing
(k-)Set Agreement 285
Martin Biely, Peter Robinson, and Ulrich Schmid
Unifying Byzantine Consensus Algorithms with Weak Interactive
Consistency 300
Zarko Milosevic, Martin Hutle, and Andr´e Schiper
Distributed Algorithms
Safe and Eventually Safe: Comparing Self-stabilizing
and Non-stabilizing Algorithms on a Common Ground
(Extended Abstract) 315
Sylvie Dela¨et, Shlomi Dolev, and Olivier Peres
Proactive Fortification of Fault-Tolerant Services 330
Paul Ezhilchelvan, Dylan Clarke, Isi Mitrani, and
Santosh Shrivastava
Robustness of the Rotor-router Mechanism 345
Evangelos Bampas, Leszek G¸asieniec, Ralf Klasing,
Adrian Kosowski, and Tomasz Radzik
Brief Annoucement: Analysis of an Optimal Bit Complexity
Randomised Distributed Vertex Colouring Algorithm
(Extended Abstract) 359
Yves M´etivier, John Michael Robson, Nasser Saheb-Djahromi, and
Akka Zemmari
Brief Annoucement: Distributed Swap Edges Computation for
Minimum Routing Cost Spanning Trees 365
Linda Pagli and Giuseppe Prencipe
Author Index 373
[...]... independent of Nmax Even though our bounding analysis has a complexity which is exponential in the number of processors, we observe that multiprocessor embedded systems are usually composed of a limited number of processors In our context, this makes our bounds efficient where L L VNk = E[(TNk )2 ] = Homogeneous Processors In many cases of practical interest, multiprocessor systems are composed of identical... moments of T by means of Theorem 2 In turn, the mean waiting W straightforwardly becomes a lower bound by means of (7) In (9), the computational complexity of T L and V L is then dominated by the computation of T1 (note that V1 is obtained at the same computational cost as T1 ) By means of Formula (4), this is given by O(R2R + K) for time and O(R+K) for space Therefore, the computational complexity of the... current state of each processor (idle or not) Second, it is asymptotically optimal in terms of worst-case complexity [6] Finally, it is processor oblivious since it automatically adapts on-line to the number and the size of jobs in the system as well as to the changing speeds of processors [8] Many variants of work stealing have been developed In the following, we will consider a special case of the work-stealing... among processors The goodness of these strategies turns out to strongly depend on the structure of communication costs so that their impact is non-trivial to predict without our model Due to space limitations, we refer to [4] for proofs, details and additional experimental results 2 Model of Work Stealing over a Multi-processor Architecture To assess the performance of the systems introduced above, one... discipline of jobs is FCFS and their service time in processor r is exponential with rate μ−1 During the execution of task i, if r processor r becomes idle, then it attempts to steal nmax /2 jobs from the queue of the processor with the largest number of jobs, i.e., nmax When a processor steals jobs from the queue of another processor, it uses the communication bus Performance Evaluation of Work Stealing... Proof The above formulas are obtained by applying standard one-step analysis and taking into account the transition rates in Figure 1 of the Markov chain characterizing the task service time distribution For more details on the interpretation of the formula above see [4] We now explicit performance indices formulas of the proposed work-stealing model which are expressed in terms of the results of Theorem... ,Nmax requires the computation of TNk , ,Nk , for all k, the direct computation of T through (6) has the Performance Evaluation of Work Stealing for Streaming Applications 25 complexity of computing TNmax , ,Nmax Assuming that one can iterate over set Ω(i) := {n : r nr = i, 0 ≤ nr ≤ Nmax } in O( Ω(i) ) steps (by means, e.g., of recursive calls), the computational requirements of the proposed analysis beR... space truncation of process (1) which limits to M the number of tasks in the system For a given λ, it is known that such truncation yields nearly exact results if M is sufficiently large (note that M should be much larger than RNmax ) The resulting complexity is given by the computational R requirement of the solution of a linear system composed of O(M Nmax ) equations, which is orders of magnitude worse... positive is enqueued in the head of the work dequeue of the core where it executed 5.2 Analysis of the B IMODAL Scheduler We first bound (from below) the makespan that can be achieved by an optimal conservative scheduler Theorem 5 For every workload Γ , the makespan of Γ under an optimal, conservative τ of ine scheduler O PT satisfies makespanOpt (Γ ) ≥ max{ sωi , m i } Proof There are m cores, and hence,... Proof Consider the scheduling of a bimodal workload Γ under B IMODAL Let tk be the starting time of the last reading epoch after all the work deques of cores are empty, and such that some transactions arrive after tk At time tk , no transactions are available in the work queues of any core, and hence, no matter what the optimal scheduler O PT does, its makespan is at least tk Let Γk be the set of . Conference on Principles of Distributed Systems, is
an annual forum for presentation of state -of- the-art knowledge on principles of
distributed computing systems, . related
to principles of distributed computing and distributed systems. We hope that this
13th edition will contribute to the growth and the development of the
Ngày đăng: 22/03/2014, 09:20
Xem thêm: Principles of Distributed Systems pot, Principles of Distributed Systems pot