Distributed scheduling for multicomputers
Load sharing or load balancing are generally described as ways to get better execution performances: lots of experimentations have been successfully carried out, usually on networks of workstations. In this article we present new aspects of the load balancing issue when applying it to multicomputers. The main constraints and motivations for a load balancing service on multicomputers are studied. We show that this service could rise to a distributed scheduling service, integrated in the operating system and could avoid the hard process mapping task. A description of a load balancing implementation under Chorus/MiX is given.
KeywordsLoad Balance Process Migration Load Sharing Displacement Policy Load Manager
Unable to display preview. Download preview PDF.
- [Barak85]A. Barak and all A Distributed Load-balancing Policy for a Multicomputer Software Practice and Experience July 85Google Scholar
- [Doug91]F. Douglis and all Transparent Process Migration: Design Alternatives and the Sprite Implementation Software Practice and Experience August 91Google Scholar
- [Eager86]D. L. Eager and all Adaptive Load Sharing in Homogeneous Distributed Systems IEEE transactions on software engineering, May 1986Google Scholar
- [Krue88]P. Krueger and all A Comparison of Preemptive and Non-Preemtive Load Distributing 8th ICDCS San Jose, CA June 88Google Scholar
- [Lela86]W. E. Leland and all Load-Balancing Heuristics and Process Behavior ACM SIGMETRICS 1986, Raleigh, NCGoogle Scholar
- [Lin86]C. H. Lin and all Gradient Model: a Demand-driven Load Balancing Scheme 6th ICDCS, Cambridge, MA May, 86Google Scholar
- [Phil93]Laurent Philippe Contribution à l'étude et á la réalisation d'un système á image unique pour multicalculateur Phd Thesis University of Franche-Comté Janurary 93Google Scholar
- [Thei85]M. Theimer and all Preemptive Remote Execution Facilities for the V System ACM Press: Operating System review December 1985Google Scholar
- [Zaya87]E. Zayas Attacking the Process Migration Bottleneck 11th ACM SOSP, Austin, Texas November 1987Google Scholar