Abstract
A key issue of dynamic load balancing in a loosely coupled distributed system is selecting appropriate jobs to transfer. In this paper, a job selection policy based on on-line predicting behaviors of jobs is proposed. Tracing is used at the beginning of execution of a job to predict the approximate execution time and resource requirements of the job so as to make a correct decision about whether transferring the job is worthwhile. A dynamic load balancer using the job selection policy has been implemented. Experimental measurement results show that the policy proposed is able to improve mean response time of jobs and resource utilization of systems substantially.
Similar content being viewed by others
References
Douglis F, Ousterhoat J. Transparent process migration: Design alternatives and the sprite implementation.Software-Practice and Experience, 1991, 21(8): 757–785.
Krueger P, Livny M. A comparison of preemptive and non-preemptive load distributing. InProc. 8th Int. Conf. Distributed Comput. Syst., pp. 123–130, June 1988.
Wang C, Krueger P, Liu M T. Intelligent job selection for distributed scheduling. InProc. 13th Int. Conf. Distributed Comput. Syst., pp.517–524, May 1993.
Eager D, Lazowska E, Zahorjan, J. A dynamic load sharing in homogeneous distributed systems.IEEE Trans. on Software Eng., 1986, SE-12(5): 662–675.
Leland W, Ott T. Load balancing heuristics and process behavior. InProc. ACM SIGMETRICS Conf. Measurement and Modeling of Computer Syst., May 1986.
Cabrera L M. The influence of workload on load balancing strategies. InProc. Summer USENIX Conf., pp. 446–458, June 1986.
Zhou S. A trace-driven simulation study of dynamic load balancing.IEEE Trans. on Software Eng., 1988, SE-14(9): 1327–1341.
Svensson A. History, an intelligent load sharing filter. InProc. 10th Int. Conf. Distributed Comput. Syst., May 1990, pp.546–553.
Goswami K K, Devarakonda M, Iyer R K. Prediction-based dynamic load-sharing heuristics.IEEE Trans on Parallel and Distributed Syst., 1993, 4(6): 638–648.
Devarakonda M, Iyer R K. Predictability of process resource usage: A measurement-based study of UNIX.IEEE Trans. on Software Eng., 1989, SE-15(12).
Ezzat A K. Load balancing in NEST: A network of workstation. InProc. Fall Joint Computer Conf., Nov. 1986.
Zhou S, Ferrari D. A measurement study of load balancing performance. InProc. 7th Int. Conf. Distributed Comput. Syst., pp.490–497, May 1987.
Baumgartner J M, Wha B W. GAMMON: A load balancing strategy for local computer systems with multiaccess networks.IEEE Trans. on Computers, 1989, C-38(8): 1098–1109.
Koch T, Rohde G, Kramer B. Adaptive load balancing in a distributed environment. InProc. The First Workshop on Services of Distributed and Networked Environments, pp.115–121, June 1994.
Ju J, Xu G, Tao J. Parallel computing using idle workstations.ACM Operating Systems Review, 1993, 27(3): 87–96.
Author information
Authors and Affiliations
Additional information
This work was supported by the National High-Tech R/D Programme and the National Natural Science Foundation of China.
Ju Jiubin is a Professor in the Department of Computer Science, Jilin University. His research interests include parallel processing, computer networks, computer architecture, and distributed systems.
Xu Gaochao is a Teacher in the Department of Computer Science, Jilin University. He received his B.S., M.S. and Ph.D. degrees in computer science from Jilin University in 1988, 1991, and 1995, respectively. His research interests include parallel processing, computer networks, and distributed systems.
Yang Kun received his M.S. degree in computer science from Jilin University in 1994. His research interests include parallel processing, computer networks, and distributed systems.
Rights and permissions
About this article
Cite this article
Ju, J., Xu, G. & Yang, K. On-line predicting behaviors of jobs in dynamic load balancing. J. of Compt. Sci. & Technol. 11, 39–48 (1996). https://doi.org/10.1007/BF02943520
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF02943520