This Article 
 Bibliographic References 
 Add to: 
A Trace-Driven Simulation Study of Dynamic Load Balancing
September 1988 (vol. 14 no. 9)
pp. 1327-1341

A trace-driven simulation study of dynamic load balancing in homogeneous distributed systems supporting broadcasting is presented. Information about job CPU and input/output (I/O) demands collected from production systems is used as input to a simulation model that includes a representative CPU scheduling policy and considers the message exchange and job transfer cost explicitly. Seven load-balancing algorithms are simulated and their performances compared. Load balancing is capable of significantly reducing the mean and standard deviation of job response times, especially under heavy load, and for jobs with high resource demands. Algorithms based on periodic or nonperiodic load information exchange provide similar performance, and, among the periodic policies, the algorithms that use a distinguished agent to collect and distribute load information cut down the overhead and scale better. With initial job placements only, source initiative algorithms were found to perform better than server initiative algorithms. The performances of all hosts, even those originally with light loads, are generally improved by load balancing.

[1] A. Barak and A. Shiloh, "A distributed load balancing policy for a multicomputer," Dep. Comput. Sci., Hebrew Univ. of Jerusalem, 1984.
[2] B.N. Bershad, "Load Balancing With Maitre d'," Computer Science Division UCB/CSD 86/276, Univ. of Calif., Berkeley, Dec. 1985.
[3] S. H. Bokhari, "Dual processor scheduling with dynamic reassignment,"IEEE Trans. Software Eng., vol. SE-5, no. 4, pp. 341-349, July 1979.
[4] R. Bryant and R. Finkel, "A stable distributed scheduling algorithm," inProc. Int. Conf. Distributed Processing Systems, 1981, pp. 314-323.
[5] L. F. Cabrera, E. Hunter, M. Karels, and D. Mosher, "A user-process oriented performance study of Ethernet networking under Berkeley UNIX 4.2 BSD," to appear inIEEE Trans. Software Eng.; also as Tech. Rep. UCB/CSD 84/216, Comput. Sci. Division, Univ. California, Berkeley, Dec. 1984.
[6] L. F. Cabrera, "The influence of workload on load balancing strategies," inProc. 1986 Summer USENIX Conference, Atlanta, GA, June 1986, pp. 446-458.
[7] Y. Chow and W. Kohler, "Models of dynamic load balancing in a heterogeneous multiple processor system,"IEEE Trans. Comput., vol. C-28, no. 5, pp. 354-361, May 1979.
[8] D. Eager, E. Lazowska, and J. Zahorjan, "A comparison of receiverinitiated and sender-initiated dynamic load sharing," Dep. Comput. Sci. Univ. Washington, Tech. Rep. 85-04-01, Apr. 1985.
[9] 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.
[10] D. Ferrari,Computer Systems Performance Evaluation. Englewood Cliffs, NJ: Prentice-Hall, 1978.
[11] D. Ferrari, "A study of load indices for load balancing schemes," Comput. Sci. Division, Univ. California, Berkeley, Tech. Rep. UCB/CSD 85/ 262, Oct. 1985; also inWorkload Characterization of Computer Networks, North-Holland, 1986.
[12] A. Hac and T. J. Johnson, "A study of dynamic load balancing in a distributed system," G. Serazzi, Ed. Amsterdam, The Netherlands: inProc. ACM SIGCOMM Symp. Communications, Architectures and Protocols, Stowe, VT, Aug. 1986, pp. 348-356.
[13] R. Hagmann, "Process server: Sharing processing power in a workstation environment," inProc. Principles of Distributed Computing, Cambridge, MA, May 1986.
[14] K. Hwang, W. Croft, G. Goble, B. Wah, F. Briggs, W. Simmons, and C. Coates, "A UNIX-based local computer network with load balancing,"Computer, vol. 15, no. 4, pp. 55-66, Apr. 1982.
[15] E. Lazowska, J. Zahorjan, D. Cheriton, and W. Zwaenepoel, "File access performance of diskless workstations," Dep. Comput. Sci., Univ. of Washington, Tech. Rep. 84-06-01, June 1984.
[16] W. E. Leland and T. J. Ott, "Load-balancing heuristics and process behavior," inProc. 1986 ACM Sigmetrics Conf. Measurement Modeling Comput. Syst. Perform. Eval. Rev., vol. 14, no. 1, pp. 54-69, North Carolina State Univ., May 1986.
[17] M. Livny and M. Melmen, "Load balancing in homogeneous broadcast distributed systems," inProc. Computer Network Perform. Symp., 1982, pp. 47-55.
[18] H. S. Stone, "Multiprocessor scheduling with the aid of network flow algorithms,"IEEE Trans. Software Eng., vol. SE-3, no. 1, pp. 85- 93, Jan. 1977.
[19] H. S. Stone, "Critical load factors in two processor distributed systems,"IEEE Trans. Software Eng., vol. SE-4, no. 3, pp. 254-258, May 1978.
[20] M. Theimer, K. Lantz, and D. Cheriton, "Preemptive remote execution facilities for the V-System," Dep. Comput. Sci., Stanford Univ., Tech. Rep. STAN-CS-85-1087, Sept. 1985.
[21] Y. Wang and R. Morris, "Load balancing in distributed systems,"IEEE Trans. Comput., C-34, no. 3, pp. 204-217, Mar. 1985.
[22] S. Wu and M. Liu, "Assignment of tasks and resources for distributed processing," inProc. COMPCON, Fall 1980, pp. 655-662.
[23] S. Zhou, "An experimental assessment of resource queue lengths as load indices," inProc. Winter USENIX Conf., Washington, DC, Jan. 1987, pp. 73-82; also as Tech Rept. UCB/CSD 85/298, Comput. Sci. Division, Univ. Calif., Berkeley, Apr. 1986.
[24] S. Zhou and D. Ferrari, "An experimental study of load balancing performance," inProc. 7th Int. Conf. Dist. Computing Syst., Berlin, Sept. 1987, pp. 490-497; also as Tech. Rep. UCB/CSD 87/336, Comput. Sci. Division, Univ. Calif., Berkeley, Jan. 1987.
[25] S. Zhou, "Performance studies of dynamic load balancing in distributed systems," Ph.D. dissert., Univ. California, Berkeley, Sept. 1987.

Index Terms:
trace-driven simulation; dynamic load balancing; homogeneous distributed systems; broadcasting; CPU; input/output; scheduling; message exchange; job transfer; job response times; distributed databases; input-output programs; scheduling
S. Zhou, "A Trace-Driven Simulation Study of Dynamic Load Balancing," IEEE Transactions on Software Engineering, vol. 14, no. 9, pp. 1327-1341, Sept. 1988, doi:10.1109/32.6176
Usage of this product signifies your acceptance of the Terms of Use.