Advertisement

A comparative analysis of two-phase-commit protocols

  • Ouri Wolfson
Fault Tolerance
Part of the Lecture Notes in Computer Science book series (LNCS, volume 470)

Abstract

We compare the performance of four variants of the two-phase-commit paradigm, in the absence of failures. They are tree-commit ([SW1]), the decentralized ([Sk]), the linear ([G]), and the central-site ([LS]) algorithms. The performance measures are communication cost and communication time. We find that the communication cost of tree-commit is equal to that of the linear and central-site algorithms, its communication time cannot be worse, but it can be twice as fast. The communication time of the decentralized algorithm is better than that of tree-commit, whereas tree-commit wins as far as communication cost is concerned. When comparing the communication-complexity of the two algorithms, namely the product of the communication cost and the communication time, tree-commit wins.

Keywords

Communication Cost Minimum Span Tree Communication Complexity Communication Time Consensus Problem 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [BHG]
    P. Bernstein, V. Hadzilacos, and N. Goodman, Concurrency Control and Recovery in Database Systems Addison Wesley, 1987.Google Scholar
  2. [CP]
    S. Ceri and G. Pelagatti, Distributed Database Principles and Systems, McGraw-Hill, 1984.Google Scholar
  3. [DS1]
    C. Dwork and D. Skeen, "The Inherent Cost of Nonblocking Commitment", Prod. 2nd ACM Symp. on PODC, pp. 1–11, 1983.Google Scholar
  4. [DS2]
    C. Dwork and D. Skeen, "Patterns of Communication in Consensus Protocols", Proc. 3rd ACM Symp. on PODC, pp. 143–153, 1984.Google Scholar
  5. [E]
    S. Even, Graph Algorithms, Computer Science Press, 1979.Google Scholar
  6. [F]
    M. Fischer, The Consensus Problem in Unreliable Distributed Systems (a brief survey), Technical Report YALEU/DCS-/RR-273, Yale University, June 1983.Google Scholar
  7. [G]
    J.N. Gray, "Notes on Database Operating Systems," Operating Systems: An Advanced Course, Springer-Verlag, 1979.Google Scholar
  8. [H1]
    V. Hadzilacos, "On the Relationship between the Atomic Commitment and Consensus Problems," Proc. of the Workshop on Fault-Tolerant Distributed Computing Springer Verlag, 1986.Google Scholar
  9. [H2]
    V. Hadzilacos, "A Knowledge Theoretic Analysis of Atomic Commitment Protocols," Proc. 6th ACM Symp. on PODS, pp. 129–134, 1987. A revised version has been submitted for publication.Google Scholar
  10. [HHL]
    S. Hadetniemi, S. Hadetniemi, and A. Liestman, "A Survey of Gossiping and Broadcasting in Communication Networks," Networks Vol. 18 pp. 319–349, 1988.Google Scholar
  11. [I]
    A. Itai, Unpublished result, 1986.Google Scholar
  12. [L]
    L. Lamport, "Time, Clocks, and the Ordering of Events in a Distributed System," CACM, 21(7), pp. 558–565, 1978.Google Scholar
  13. [LHJLSW]
    B. Liskov, M. Herlihy, P. Johnson, G. Leavens, R. Scheifler, and W. Weihl, "Preliminary Argus Reference Manual", Programming Methodology Group Memo 39, 1983.Google Scholar
  14. [LS]
    B. Lampson and H. Sturgis, "Crash Recovery in a Distributed Database System," TR, Xerox PARC, 1976.Google Scholar
  15. [MLO]
    C. Mohan, B. Lindsay, and R. Obermack, "Transaction Management in the R* Distributed Database Management System," TODS, 11(4), pp. 378–396, 1986.Google Scholar
  16. [ML]
    C. Mohan, B. Lindsay, "Efficient Commit Protocols for the Tree of Processes Model of Distributed Transactions" Proc. 2nd ACM Symp. on PODC, pp. 76–88, 1983.Google Scholar
  17. [R]
    K.V.S. Ramarao, "On the Complexity of Commit Protocols," Proc. 4th ACM Symp. on PODS, pp. 235–244, 1985.Google Scholar
  18. [Sk]
    D. Skeen, "Nonblocking Commit Protocols," Proc. ACM SIGMOD, pp. 133–142, 1981.Google Scholar
  19. [Sp]
    A. Spector, "Modular Architectures for Distributed and Database Systems," Proc 8th ACM Symp. on PODS, pp. 217–224, 1989.Google Scholar
  20. [SW1]
    A. Segall, O. Wolfson, "Transaction Commitment at Minimal Communication Cost", Proceedings of the 6th ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems (PODS), San Diego CA, March 1987, pp. 112–118.Google Scholar
  21. [SW2]
    A. Segall, O. Wolfson, "Optimal Communication Topologies for Atomic Commitment" Proceedings of the Fourth International Conference on Data Engineering (DE4), Los Angeles, CA, February 1988, pp. 51–57.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1990

Authors and Affiliations

  • Ouri Wolfson
    • 1
  1. 1.Department of Computer ScienceColumbia UniversityNew York

Personalised recommendations