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