In recent years, a dramatic rise in the number of personal workstations interconnected via local area networks has occurred in the workplace. These can be organized as distributed computing systems. The combined computing power of these systems are often greater than mainframes of a decade ago, and usually less expensive. There is a growing interest in harnessing this often-underutilized power. Researchers are focusing their attention on remote execution of processes as one solution. An additional topic of research is to balance a workload among a series of computers. Remote execution is made possible because the distributed operating system provides migration of a process from one processor to another. The goals of dynamic process migration are to be user transparent, network topology independent, and to achieve a useful compromise between maximum processor utilization and minimum communication overhead. Two load sharing algorithms have been proposed as the "better" solution. The drafting algorithm is receiver initiated by a low load processor while the bidding algorithm is sender initiated by a high load processor. To date there has been no known published comparison of implemented systems. The purpose of this paper is to compare the performance of each for a real implementation rather than a paper study.
A. Ross and B. M. McMillin, "Experimental Comparison Of Bidding And Drafting Load Sharing Protocols," Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990, vol. 2, pp. 968 - 974, article no. 556306, Institute of Electrical and Electronics Engineers, Jan 1990.
The definitive version is available at https://doi.org/10.1109/DMCC.1990.556306
International Standard Book Number (ISBN)
Article - Conference proceedings
© 2023 Institute of Electrical and Electronics Engineers, All rights reserved.
01 Jan 1990