|
Article information
2017 , Volume 22, Special issue, p.4-12
Baranov A.V., Tikhomirov A.I.
Scheduling of jobs in a territorially distributed computing system with absolute priorities
This study adresses research and design of approaches and algorithms for scheduling of jobs with absolute priorities and unpredictable run time in a territorially distributed computing system (TDCS). To this purpose, authors designed and researched the TDCS model comprising of several high-performance computing (HPC) clusters united by communication channels with variable bandwidth. Both local and global levels of management are reviewed in the model. On a local level, jobs go through a local cluster queue to be run on a single HPC cluster. On a global level, jobs go through the global TDCS queue with global scheduler submitting jobs to one of local cluster queues. Jobs have absolute priorities. High priority job is able to interrupt execution of a low priority job and return it to the queue. Minimizing of the staying time for high priority jobs is the goal of the global scheduler. The researched model was implemented as a prototype of the TDCS. The decentralized dispatching scheme and the scheduling algoritm were designed for the prototype. Algorithm distributes jobs to the HPC clusters considering performance, workload of clusters that consider and bandwidth of cluster communication channels were presented. The pilot operation of the prototype was done successfully. In authors opinion following provisions and results were obtained: for the model of TDCS with absolute prioritization and for the decentralized dispatching scheme and the algorithm of scheduling jobs with absolute priorities
[full text]
Keywords: grid, absolute priorities, resource management, running time, scheduling
Author(s):
Baranov Anton Viktorovich
PhD. , Associate Professor
Position: Leading research officer
Office: Joint Supercomputer Center of the Russian Academy of Sciences
Address: 119334, Russia, Moscow, 32a, Leninsky Prospect
Phone Office: (495) 954-02-21
E-mail: antbar@mail.ru
SPIN-code: 3595-7482Tikhomirov Artem Igorevich
Position: researcher
Office: Joint Supercomputer Center of the Russian Academy of Sciences
Address: 119334, Russia, Moscow, 32a, Leninsky Prospect
Phone Office: (495) 954-02-21
E-mail: TEMA4277@rambler.ru
References:
1. Foster, I. The anatomy of the Grid: enabling scalable virtual organizations. International Journal of High Performance Computing Applications. 2001; 15(3):200-222.
2. Foster, I. The Physiology of the Grid: An Open Grid Services Architecture for Distributed Systems Integration. Computer Networks: The International Journal of Computer and Telecommunications Networking. 2002; 40(1):5-17.
3. Kovalenko, V.N. Evolution and problems of Grid. Open systems. 2003; (1):23–33. (In Russ.)
4. Kovalenko, V.N., Kovalenko, E.I., Koryagin, D.A. Resource manager for GRID with global job queue and planning based on local schedules. Trudy Pervoy Vserossiyskoy nauchnoy konferentsii «Metody i sredstva obrabotki informatsii» [Proceedings of the First All-Russian Scientific Conference «Methods and Means of Information Processing»]. Moscow: MGU im. M.V. Lomonosova; 2003: 205–210. (In Russ.)
5. Toporkov, V.V., Emelyanov, D.M., Potekhin, P.A. Job batch generation and scheduling in distributed computing environment. Bulletin of the South Ural State University. Series: Computational mathematics and software engineering. 2015; 4(2):44–57. (In Russ.)
6. Kovalenko, V.N., Kovalenko, E.I., Shorin, O.N. Razrabotka dispetchera zadaniy grid, osnovannogo na operezhayushchem planirovanii [Development of grid job dispatcher based on lookahead scheduling]. Moscow: Institut prikladnoy matematiki imeni M.V. Keldysha; 2005: 28. (In Russ.)
7. Toporkov, V.V. Modeli raspredelennykh vychisleniy [Distributed Computing Models]. Moscow: FIZMATLIT; 2005: 320. ( In Russ.)
8. Barskiy, A.B. Parallel'nye protsessy v vychislitel'nykh sistemakh. Planirovanie i organizatsiya [Parallel Processes in Computing Systems. Planning and organization]. Moscow: Radio i svyaz'; 1990: 256. (In Russ.)
9. Dlya nachinayushchikh pol'zovateley vychislitel'nykh klasterov [For novice users of computing clusters]. Available at: https://parallel.ru/cluster/beginnerguide.html (accessed 23.09.2016). (In Russ.)
10. Mutz, A., Wolski, R., Brevik, J. Eliciting Honest Value Information in a Batch-queue Environment. Proc. of the «8th IEEE/ACM International Conference on Grid Computing». Austin: IEEE Computer Society; 2007:291 – 297
11. Ernemann, C. Economic scheduling in Grid computing. Job Scheduling Strategies for Parallel Processing. 2002; (2537):129-152.
12. Kovalenko, V.N., Semyachkin, D.A. Using BackFill in GRID system. Trudy mezhdunarodnoy konferentsii «Raspredelennye vychisleniya i Grid-tekhnologii v nauke i obrazovanii» [Processing of the «International Conference on Distributed Computing and Grid-technologies in Science and Education»]. Dubna: Ob"edinennyy institut yadernykh issledovaniy; 2004:139–144.
13. Towsley, D. Analysis of Fork-Join Program Response Times on Multiprocessors. IEEE Transactions Parallel and Distributed System. 1990; 1(3):286–303.
Bibliography link:
Baranov A.V., Tikhomirov A.I. Scheduling of jobs in a territorially distributed computing system with absolute priorities // Computational technologies. 2017. V. 22. XVII All-Russian Conference of Young Scientists on Mathematical Modeling and Information Technology. P. 4-12
|
