Juan Luis Jiménez Laredo
Frédéric Guinand
ABSTRACT
This paper aims at investigating platform-independent measures to determine the efficiency of a load-balancer. Counterpoised curves are proposed as a mean to counterweight the utilization of resources with a penalty due to the migration of tasks. As a proof-of-concept, experiments were conducted on the sandpile load-balancer showing that its maximum counterpoised efficiency is reached at 80% of the peak performance of the system.
- M. R. Garey and D. S. Johnson. Computers and Intractability; A Guide to the Theory of NP-Completeness. W. H. Freeman & Co., New York, NY, USA, 1990. Google Scholar
Digital Library
- J. Laredo, P. Bouvry, F. Guinand, B. Dorronsoro, and C. Fernandes. The sandpile scheduler. Cluster Computing, 17(2):191--204, June 2014. Google ScholarDigital Library
- J. Laredo, B. Dorronsoro, J. Pecero, P. Bouvry, J. Durillo, and C. Fernandes. Designing a self-organized approach for scheduling bag-of-tasks. In P2P, Parallel, Grid, Cloud and Internet Computing, 2012 Seventh International Conference on, pages 315--320, Nov 2012. Google ScholarDigital Library
- P. Bak, C. Tang, and K. Wiesenfeld. Self-organized criticality: An explanation of the 1/f noise. Phys. Rev. Lett., 59:381--384, Jul 1987.Google ScholarCross Ref
Index Terms
- Trading Off Resource Utilization and Task Migrations in Dynamic Load-balancing
Recommendations
Balance Resource Utilization BRU Approach for the Dynamic Load Balancing in Cloud Environment by Using AR Prediction Model
One of the major challenges for the cloud provider is the efficient utilization of the physical resources. To achieve this, this paper proposed a Balance Resource Utilization BRU approach that not only minimizes the resource leakage but also increases ...
Comments