|
|
|
Polezhaev P.N. SIMULATION RESEARCH OF JOB SCHEDULING ALGORITHMS AND DATAFLOW CONTROL METHODS FOR CLOUD BASED GRIDSThis paper presents research results of job scheduling algorithms and dataflow control methods for cloud based grids. Research was carried out by grid simulator. Paper describes proposed job scheduling algorithms Backfill SDM and Backfill MDM which were compared to classical Backfill FF. Study has showed the efficiency of Backfill SDM by performance and balancing criteria. Additionally, combinations of this algorithm with conventional and developed reactive and proactive SDN dataflow control methods were estimated by simulator which showed the efficiency of Backfill SDM and proactive control method combination. This combination can be recommended for practical usage in real computing systems — clusters and grids.Key words: job scheduling, cloud computing, software defined networks, dataflow control methods.
References:
1. Feitelson, D. Utilization and predictability in scheduling the IBM SP2 with backfilling / D. Feitelson, A. Weil // Proceedings of the First Merged International Parallel Processing Symposium and Symposium on Parallel and Distributed Processing. — 1998. — P. 542-546.
2. Kovalenko, V.N. Using Backfill algorithm in grids / V.N. Kovalenko, D.A. Semyachknin // Proceedings of international conference " Raspredelennye vychislenija i Grid-tehnologii v nauke i obrazovanii ". — 2004. — P. 139-144.
3. Polezhaev, P.N. Simulator of computing cluster and its control system used for research of job scheduling algorithms // Vestnik UUrGU. "Matematicheskoe modelirovanie i programmirovanie ", 2010. — №35(211), issue. 6. — P. 79 — 90.
4. McKeown N., Anderson T., Balakrishnan H., Parulkar G., Peterson L., Rexford J., Shenker S., Turner J.; Openflow: enabling innovation in campus networks // ACM SIGCOMM Computer Communication Review, 2008, vol. 38, p. 69-74.
5. Polezhaev, P.N. Mathematical model of distributed computing datacenter consisted of software defined networks of its segments // Vestnik " Orenburgskogo gosudarstvennogo universiteta ", 2013. — 5(154) — P. 198-204.
6. Polezhaev P.N., Ushakov U.A., Shukhman A.E. Resource control system for high performance computing based on software defined networks. // Sistemy upravlenija i informacionnye tehnologii, №4(54), 2013. — P. 65-69
7. Lublin, U. The workload on parallel supercomputers: modeling the characteristics of rigid job [Текст] / U. Lublin, G. Feitelson //Journal of Parallel and Distributed Computing archive. — 2003. –Vol. 63, issue 11. — P. 542-546
About this article
Author: Polezhaev P.N.
Year: 2014
|
|
Editor-in-chief |
Sergey Aleksandrovich MIROSHNIKOV |
|
|