Skip to main content
SpringerLink
Log in

Optimal Randomized Scheduling by Replacement

  • Published:
Journal of Combinatorial Optimization Aims and scope Submit manuscript

Abstract

In the replacement scheduling problem, a system consists of n processors drawn from a pool of p,all initially “alive”. At any time some processor can die. The scheduler is immediately informed of the fault butnot of its location. It must then choose another set of n processors. If this new set contains a dead processor, the system crashes and halts. The performance of a scheduling protocol is measured by the expected number of deaths the system tolerates before it crashes. We provide an optimal randomized scheduling protocol for this problem.

The framework of this work combines an absolute performance measure for protocols and so-called adaptive online adversaries. This framework is rarely addressed because of the complexity of the interaction between protocols and adversaries. A major contribution of ourwork is to provide a theoretical foundation for the analysis of this interaction. In particular we make explicit how the protocol and the adversary affect the probability distribution of the analysis—a very general problem. We carefully analyze the exchange of sinformation between the two players, and reveal how they use their information optimally. The optimality of the protocol is established by using of a saddle point method for protocols and adversaries.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

References

  • A. Abrahmson, A. Adler, L. Higham, and D. Kirkpatrick, “Tight lower bounds for probabilistic solitude verification on anonymous rings,” J. Assoc. Comput. Mach., vol. 41, no.2, pp. 277–310, 1994.

    Google Scholar 

  • J.P. Aubin, “Mathematical methods of game and economic theory,” Studies Mathematics and its Applications, North-Holland, 1982.

  • S. Ben-David, A Borodin, R.M. Karp, G. Tardos, and A. Widgerson, “On the power of randomization in on-line algorithms,” Algorithmica, vol. 11, no.1, pp. 2–14, 1994.

    Google Scholar 

  • A. Fiat, Y. Rabani, and Y. Ravid, “Competitive k-Server Algorithms,” Proceedings of the 31st Annual IEEE Symposium on Foundations of Computer Science, 1990, pp. 454–463.

  • M. Goemans, N. Lynch, and I. Saias, “Upper and lower bounds on the number of faults a system can withstand without repairs,” Dependable Computing and Fault-Tolerant Systems 9, Springer Verlag, 1995.

  • R. Graham and A. Yao, “On the Improbability of Reaching Byzantine Agreements,” Proceedings of the ACM Symposium on Theory of Computing, 1989, pp. 467–478.

  • D. Grigoriev, M. Karpinski, F. Meyer auf der Heide, and R. Smolensky, “ALower Bound for Randomized Algebraic Decision Trees,” Proceedings of 28th IEEE Symposium on the Theory of Computing, 1996, pp. 612–619.

  • A. Karlin and A. Yao, “Probabilistic lower bounds for byzantine agreement,” unpublished document, 1984.

  • E. Kushilevitz, Y. Mansour, M. Rabin, and D. Zuckerman, “Lower Bounds for Randomized Mutual Exclusion,” Proceedings of the 25th ACM Symposium on Theory of Computing, 1993, pp. 154–163.

  • M.S. Manasse, L.A. McGeoch, and D.D. Sleator, “Competitive algorithm for server problems,” Journal of Algorithms, vol. 11, pp. 208–230, 1990.

    Google Scholar 

  • L.A. McGeoch and D.D. Sleator, “A strongly competitive randomized paging algorithm,” Algorithmica, vol. 6, pp. 816–825, 1991.

    Google Scholar 

  • R. Motwani and P. Raghavan, Randomized Algorithms, Cambridge University Press, 1995.

  • M. Rabin, “N-process mutual exclusion with bounded waiting by log N-shared variables,” J. of Computer and System Sciences, vol. 25, pp. 66–75, 1982.

    Google Scholar 

  • I. Saias, “Proving Probabilistic Correctness: The Case of Rabin's Algorithm for Mutual Exclusion,” Proceedings of the 11th Annual ACM Symposium on Principles of Distributed Computing, 1992.

  • I. Saias, “Randomness versus non-determinism in randomized computing,” Ph.D. Thesis, Technical report MIT/LCS/TR-651, Laboratory for Computer Science, Massachusetts Institute of Technology, 1994.

  • M. Vardi, “Automatic Verification of Probabilistic Concurrent Finite-State Programs,” Proceedings of 26th IEEE Symposium on Foundations of Computer Science, pp. 327–338, 1985.

  • N.N. Vorob’ev, “Game theory,” Applications of Mathematics 7, Springer Verlag, 1977.

  • A. Yao, “Some Questions Related to Distributive Computing,” Proceedings of the ACM Symposium on Theory of Computing, 1979, vol. 11, pp. 209–213.

    Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Los Alamos National Laboratory, Los Alamos, New Mexico

    Isaac Saias

Authors
  1. Isaac Saias
    View author publications

    You can also search for this author in PubMed Google Scholar

Rights and permissions

Reprints and permissions

About this article

Cite this article

Saias, I. Optimal Randomized Scheduling by Replacement. Journal of Combinatorial Optimization 1, 427–456 (1998). https://doi.org/10.1023/A:1009703014438

Download citation

  • Issue Date:

  • DOI: https://doi.org/10.1023/A:1009703014438

Keywords

Search

Navigation