This Article 
 Bibliographic References 
 Add to: 
Adaptive Location Policies for Global Scheduling
June 1994 (vol. 20 no. 6)
pp. 432-444

Two important components of a global scheduling algorithm are its transfer policy and its location policy. While the transfer policy determines whether a task should be transferred, the location policy determines where it should be transferred. Based on their location policies, global scheduling algorithms can be broadly classified as receiver-initiated, sender-initiated, or symmetrically-initiated. The relative performance of these classes of algorithms has been shown to depend on the system workload. We present two adaptive location policies for global scheduling in distributed systems. These location policies are general, and can be used in conjunction with many existing transfer policies. By adapting to the system workload, the proposed policies capture the advantages of both sender-initiated and receiver-initiated policies. In addition, by adaptively directing their search activities toward the nodes that are most likely to be suitable counterparts in task transfers, the proposed policies provide short transfer latency and low overhead, and more important, high probability of finding a suitable counterpart if one exists. These properties allow these policies to deliver good performance over a very wide range of system operating conditions. The proposed policies are compared with nonadaptive policies, and are shown to considerably improve performance and to avoid causing system instability.

[1] Y. Artsy and R. Finkel, "Designing a process migration facility: the Charlotte experience,"IEEE Comput., vol. 22, no. 9, pp. 47-56, Sept. 1989.
[2] R. M. Bryant and R. A. Finkel, "A stable distributed scheduling algorithm," inProc. 2nd Int. Conf. Distrib. Comput. Syst., Apr. 1981, pp. 314-323.
[3] T. L. Casavant and J. G. Kuhl, "A taxonomy of scheduling in general-purpose distributed computing systems,"IEEE Trans. on Software Eng., vol. 14, no. 2, pp. 141-154, Feb. 1988.
[4] F. Douglis and J. Ousterhout, "Transparent Process Migration: Design Alternatives and the Sprite Implementation,"Software -- Practice and Experience, Vol. 21, No. 8, Aug. 1991, pp. 757-785.
[5] D. L. Eager, E. Lazowska, and J. Zahorjan, "A comparison of receiver-initiated and sender-initiated adaptive loading,"Perform. Eval., vol. 6, 1986.
[6] D. Eager, E. Lazowska, and J. Zahorjan, "Adaptive load sharing in homogeneous distributed systems,"IEEE Trans. Software Eng., vol. SE-12, no. 5, pp. 662-675, May 1986.
[7] M. R. Eskicioglu, "Process migration: An annotated bibliography."Newsletter, IEEE Comput. Soc. Tech. Committee Operat. Syst. and Applic. Envir., vol. 4, no. 4, pp. 5-16, Winter 1990.
[8] L. Kleinrock,Queueing Systems. Vol. 2, Theory. New York: John Wiley&Sons, 1976.
[9] P. Krueger, "Distributed scheduling for a changing environment," Tech. Rep. 780, Ph.D. dissertation, Univ. of Wis.-Madison, 1988.
[10] P. Krueger and R. Chawla, "The Stealth Distributed Scheduler,"Proc. 11th Int'l Conf. Distributed Computing Systems, IEEE CS Press, Los Alamitos, Calif., Order No. 2144, 1991, pp. 336-343.
[11] P. Krueger and M. Livny, "A comparison of preemptive and nonpreemptive load distributing," inProc. 8th Int. Conf. Distributed Comput. Syst., San Jose, June 1988, pp. 123-130.
[12] S.S. Lavenberg,Computer Performance Modeling Handbook, Academic Press, New York, 1983.
[13] E. Lazowaka, J. Zahorjan, D. Cheriton, and W. Zwaenepoel, "File access performance of diskless workstations,"ACM Trans. Comput. Syst., vol. 4, no. 3, Aug. 1986.
[14] M. J. Litzkow, M. Levy, and M. W. Mutka, "Condor--a hunter of idle workstations," inProc. IEEE Int. Conf. on Distributed Comput. Syst., 1988, pp. 104-111.
[15] M. Livny and M. Melmen, "Load balancing in homogeneous broadcast distributed systems," inProc. Computer Network Perform. Symp., 1982, pp. 47-55.
[16] M. Mutka and M. Livny, "Profiling workstation's available capacity for remote execution," inProc. of PERFORMANCE '87, Dec. 1987, pp. 529-543.
[17] M. L. Powell and B. P. Miller, "Process migration in DEMOS/MP," inProc. Ninth Symp. Oper. Syst. Principles, Bretton Woods, NH, Oct. 1983, pp. 110-119.
[18] K. Ramamritham and J. A. Stankovic, "Dynamic task scheduling in distributed hard real-time systems,"IEEE Software, vol. 1, no. 3, pp. 65-75, July 1984.
[19] H. G. Rotithor and S. S. Pyo, "Decentralized decision making in adaptive task sharing, " inProc. 2nd IEEE Symp. Parallel and Distr. Processing, Dec. 1990, pp. 34-41.
[20] M. Stumm, "The Design and Implementation of a Decentralized Scheduling Facility for a Workstation Cluster,"Proc. Second Conf. Computer Workstations, IEEE CS Press, Los Alamitos, Calif., Order No. 810, 1988, pp. 12-22.
[21] M. Theimer, K. Lantz, and D. Cheriton, "Preemptable Remote Execution Facilities for the V-System,"Proc. 10th Symp. Operating Syst. Principles, Dec. 1985, pp. 2-12.
[22] S. Zhou, "A trace-driven simulation study of dynamic load balancing,"IEEE Trans. Software Eng., vol. 14, no. 9, pp. 1327-1341, Sept. 1988.

Index Terms:
scheduling; distributed processing; resource allocation; performance evaluation; adaptive location policies; global scheduling algorithm; transfer policy; location policy; location policies; receiver-initiated; sender-initiated; symmetrically-initiated; system workload; distributed systems; search activities; short transfer latency; low overhead; probability; nonadaptive policies; system instability; distributed scheduling; load sharing; load balancing; task migration
P. Krueger, N.G. Shivaratri, "Adaptive Location Policies for Global Scheduling," IEEE Transactions on Software Engineering, vol. 20, no. 6, pp. 432-444, June 1994, doi:10.1109/32.295892
Usage of this product signifies your acceptance of the Terms of Use.