Performance evaluation of the Estimated Response Time strategy: tools, methods and an experiment

Cecile Germain-Renaud 1, 2 Alain Cady 2
1 TAO - Machine Learning and Optimisation
CNRS - Centre National de la Recherche Scientifique : UMR8623, Inria Saclay - Ile de France, UP11 - Université Paris-Sud - Paris 11, LRI - Laboratoire de Recherche en Informatique
Abstract : An extensive body of research focuses on economic and intelligent scheduling models. Conversely, the gLite matchmaking engine adopts an agnostic approach for estimating the waiting time of incoming jobs, derived from the Copernican principle: "this job is not special". An open question is the evaluation of this minimalist strategy. This work reports on the creation of the software tools required for this evaluation, the methodology and criteria, and presents preliminary results using the Grid Observatory logs of the Information System (IS) and PBS scheduler of GRIF/LAL.
Complete list of metadatas

https://hal.inria.fr/inria-00544441
Contributor : Cecile Germain <>
Submitted on : Wednesday, December 8, 2010 - 10:36:53 AM
Last modification on : Thursday, April 5, 2018 - 12:30:12 PM

Identifiers

  • HAL Id : inria-00544441, version 1

Collections

Citation

Cecile Germain-Renaud, Alain Cady. Performance evaluation of the Estimated Response Time strategy: tools, methods and an experiment. EGEE User Forum, Apr 2010, Uppsala, Sweden. ⟨inria-00544441⟩

Share

Metrics

Record views

127