Evaluation of replication and rescheduling heuristics for grid systems with varying resource availability
- Author
- Maria Chtepen (UGent) , Bart Dhoedt (UGent) , Filip De Turck (UGent) , Piet Demeester (UGent) , Filip Claeys and Peter Vanrolleghem (UGent)
- Organization
- Abstract
- As grids typically consist of heterogeneously managed subsystems with strongly varying resources, resource availability should be taken into account in the job scheduling process. This paper introduces several dynamic online scheduling heuristics that reduce task loss and execution delay resulting from resource failures. ne heuristics are based upon task replication and rescheduling of failed tasks. Characteristic to the proposed methods is the relative simplicity and the efficiency with which they are dealing with dynamic grid environments. For tuning and evaluation of the algorithms, a discrete-event simulation framework was used. Grid systems with high and low system load. as well as varying failure patterns were investigated. The experiments have shown that the proposed failure detection based heuristics provide for almost lossless task execution but can decrease system performance. while replication-based algorithms generally result in good throughput on unreliable non-excessively loaded grids, however without giving a guarantee on the number of jobs lost.
- Keywords
- task replication, dynamic scheduling, grid computing, failure detection
Downloads
-
(...).pdf
- full text
- |
- UGent only
- |
- |
- 3.19 MB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-406974
- MLA
- Chtepen, Maria, et al. “Evaluation of Replication and Rescheduling Heuristics for Grid Systems with Varying Resource Availability.” Proceedings of the 18th IASTED International Conference on Parallel and Distributed Computing and Systems, ACTA Press Anaheim, 2006, pp. 622–27.
- APA
- Chtepen, M., Dhoedt, B., De Turck, F., Demeester, P., Claeys, F., & Vanrolleghem, P. (2006). Evaluation of replication and rescheduling heuristics for grid systems with varying resource availability. Proceedings of the 18th IASTED International Conference on Parallel and Distributed Computing and Systems, 622–627. Anaheim, CA, USA: ACTA Press Anaheim.
- Chicago author-date
- Chtepen, Maria, Bart Dhoedt, Filip De Turck, Piet Demeester, Filip Claeys, and Peter Vanrolleghem. 2006. “Evaluation of Replication and Rescheduling Heuristics for Grid Systems with Varying Resource Availability.” In Proceedings of the 18th IASTED International Conference on Parallel and Distributed Computing and Systems, 622–27. Anaheim, CA, USA: ACTA Press Anaheim.
- Chicago author-date (all authors)
- Chtepen, Maria, Bart Dhoedt, Filip De Turck, Piet Demeester, Filip Claeys, and Peter Vanrolleghem. 2006. “Evaluation of Replication and Rescheduling Heuristics for Grid Systems with Varying Resource Availability.” In Proceedings of the 18th IASTED International Conference on Parallel and Distributed Computing and Systems, 622–627. Anaheim, CA, USA: ACTA Press Anaheim.
- Vancouver
- 1.Chtepen M, Dhoedt B, De Turck F, Demeester P, Claeys F, Vanrolleghem P. Evaluation of replication and rescheduling heuristics for grid systems with varying resource availability. In: Proceedings of the 18th IASTED international conference on parallel and distributed computing and systems. Anaheim, CA, USA: ACTA Press Anaheim; 2006. p. 622–7.
- IEEE
- [1]M. Chtepen, B. Dhoedt, F. De Turck, P. Demeester, F. Claeys, and P. Vanrolleghem, “Evaluation of replication and rescheduling heuristics for grid systems with varying resource availability,” in Proceedings of the 18th IASTED international conference on parallel and distributed computing and systems, Dallas, TX, USA, 2006, pp. 622–627.
@inproceedings{406974, abstract = {{As grids typically consist of heterogeneously managed subsystems with strongly varying resources, resource availability should be taken into account in the job scheduling process. This paper introduces several dynamic online scheduling heuristics that reduce task loss and execution delay resulting from resource failures. ne heuristics are based upon task replication and rescheduling of failed tasks. Characteristic to the proposed methods is the relative simplicity and the efficiency with which they are dealing with dynamic grid environments. For tuning and evaluation of the algorithms, a discrete-event simulation framework was used. Grid systems with high and low system load. as well as varying failure patterns were investigated. The experiments have shown that the proposed failure detection based heuristics provide for almost lossless task execution but can decrease system performance. while replication-based algorithms generally result in good throughput on unreliable non-excessively loaded grids, however without giving a guarantee on the number of jobs lost.}}, author = {{Chtepen, Maria and Dhoedt, Bart and De Turck, Filip and Demeester, Piet and Claeys, Filip and Vanrolleghem, Peter}}, booktitle = {{Proceedings of the 18th IASTED international conference on parallel and distributed computing and systems}}, isbn = {{9780889866386}}, keywords = {{task replication,dynamic scheduling,grid computing,failure detection}}, language = {{eng}}, location = {{Dallas, TX, USA}}, pages = {{622--627}}, publisher = {{ACTA Press Anaheim}}, title = {{Evaluation of replication and rescheduling heuristics for grid systems with varying resource availability}}, year = {{2006}}, }