Skip to main content
SpringerLink
Log in

Scheduling parallel jobs on multicore clusters using CPU oversubscription

  • Published:
The Journal of Supercomputing Aims and scope Submit manuscript

Abstract

Job scheduling strategies in multiprocessing systems aim to minimize waiting times of jobs while satisfying user requirements in terms of number of execution units. However, the lack of flexibility in the requests leaves the scheduler a reduced margin of action for scheduling decisions. Many of such decisions consist on just moving ahead some specific jobs in the wait queue. In this work, we propose a job scheduling strategy that improves the overall performance and maximizes resource utilization by allowing jobs to adapt to variations in the load through CPU oversubscription and backfilling. The experimental evaluations include both real executions on multicore clusters and simulations of workload traces from real production systems. The results show that our strategy provides significant improvements over previous proposals like Gang Scheduling with Backfilling, especially in medium to high workloads with strong variations.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15
Fig. 16
Fig. 17
Fig. 18
Fig. 19

Similar content being viewed by others

Notes

  1. We assume that two processes are executed twice slower on a single CPU than on two separated CPUs. Different studies indicate that this time could be by far less than twice [5, 6], since computation and communication can be overlapped.

References

  1. Top500 supercomputers sites (2013) http://www.top500.org/. Accessed 7 Mar 2014

  2. Feitelson DG, Rudolph L, Schwiegelshohn U, Sevcik KC, Wong P (1997) Theory and practice in parallel job scheduling. Lect Notes Comput Sci 1291:1–34

    Google Scholar 

  3. Feitelson DG, Rudolph L (1996) Toward convergence in job schedulers for parallel supercomputers. In: Job scheduling strategies for parallel processing. Springer-Verlag, New York, pp 1–26

  4. Utrera G, Corbalán J, Labarta J (2004) Implementing malleability on MPI jobs. In: Proceedings of the 13th international conference on parallel architectures and compilation techniques, PACT ’04. IEEE Computer Society, Washington, pp 215–224

  5. Subotic V, Labarta J, Valero M (2010) Simulation environment for studying overlap of communication and computation. Performance analysis of systems and software (ISPASS). In: 2010 IEEE international symposium on White Plains, pp 115–116

  6. Utrera G, Tabik S, Corbalán J, Labarta J (2011) A job scheduling approach to reduce waiting times. In: Technical report, Technical University of Catalonia (UPC-DAC-RR-2012-1, October 2011) http://www.ac.upc.edu/app/research-reports/html/2012/1/tec_repor_092011. Accessed 18 Feb 2012

  7. Feitelson DG, Rudolph L (1992) Gang scheduling performance benefits for fine-grain synchronization. J Parallel Distrib Comput 16(4):306–318

    Article  MATH  Google Scholar 

  8. Zhang Y, Franke H, Moreira J, Sivasubramaniam A (2001) An integrated approach to parallel scheduling using gang-scheduling, backfilling, and migration. Job scheduling strategies for parallel processing. In: Feitelson D, Rudolph L (eds) Lecture notes in computer science, vol 2221. Springer, Berlin, Heidelberg, pp 133–158

    Google Scholar 

  9. Parallel workload archive (2013) http://www.cs.huji.ac.il/labs/parallel/workload/. Accessed 7 Mar 2014

  10. Buisson J, Sonmez O, Mohamed H, Lammers W, Epema D (2007) Scheduling malleable applications in multicluster systems. In: Proceedings of the IEEE international conference on cluster computing 2007, pp 372–381

  11. Cera MC, Georgiou Y, Richard O, Maillard N, Navaux POA (2010) Supporting malleability in parallel architectures with dynamic cpusets mapping and dynamic MPI. In: Proceedings of the 11th international conference on distributed computing and networking, ICDCN’10. Springer-Verlag, Berlin, Heidelberg, pp 242–257

  12. El Maghraoui K, Desell TJ, Szymanski BK, Varela CA (2007) Dynamic malleability in iterative MPI applications. In: Proceedings of the 7th IEEE international symposium on cluster computing and the grid, CCGRID ’07. IEEE Computer Society, Washington, pp 591–598

  13. Iancu C, Hofmeyr S, Zheng Y, Blagojevic F (2010) Oversubscription on multicore processors. In: 24th international parallel and distributed processing symposium (IPDPS), pp 1–11

  14. Padhye J, Dowdy LW (1996) Dynamic versus adaptive processor allocation policies for message passing parallel computers: an empirical comparison. In: Proceedings of the workshop on job scheduling strategies for parallel processing. Springer-Verlag, London, pp 224–243

  15. Cirne W, Berman F (2002) Using moldability to improve the performance of supercomputer jobs. J Parallel Distrib Comput 62:1571–1601

    Article  MATH  Google Scholar 

  16. Downey AB (1997) A model for speedup of parallel programs. In: Technical report, University of California, Berkerley

  17. Sodan AC, Jin W (2010) Backfilling with fairness and slack for parallel job scheduling. J Phys Conf Ser 256(1):012–023

    Google Scholar 

  18. Sudarsan R, Ribbens CJ (2009) Scheduling resizable parallel applications. In: International parallel and distributed processing symposium, pp 1–10

  19. McCann C, Zahorjan J (1994) Processor allocation policies for message-passing parallel computers. In: Proceedings of the 1994 ACM SIGMETRICS conference on measurement and modeling of computer systems, SIGMETRICS ’94. ACM, New York, pp 19–32

  20. NAS Parallel Benchmarks (2013) http://www.nas.nasa.gov/Resources/Software/npb.html. Accessed 7 Mar 2014

  21. Wiseman Y, Feitelson DG (2003) Paired gang scheduling. IEEE Trans Parallel Distrib Syst 14(6):581–592

    Article  Google Scholar 

  22. Arpaci-Dusseau AC (2001) Implicit coscheduling: coordinated scheduling with implicit information in distributed systems. ACM Trans Comput Syst 19:283–331

    Article  Google Scholar 

  23. Zhang Y, Sivasubramaniam A, Moreira J, Franke H (2000) A simulation-based study of scheduling mechanisms for a dynamic cluster environment. In: Proceedings of the 14th international conference on supercomputing, ICS’00. ACM, New York, pp 100–109

  24. Utrera G, Corbalán J, Labarta J (2004) Scheduling of MPI applications: self-co-scheduling. In: Proceedings of the Euro-Par 2004 conference, 31th August–3rd September 2004, Italy. Lecture notes in computer science, vol 3149, pp 238–245. Springer, New York

  25. Utrera G, Tabik S, Corbalán J, Labarta J (2012) A job scheduling approach for multi-core clusters based on virtual malleability. In: Euro-Par, pp 191–203

  26. Lifka DA (1995) The ANL/IBM SP scheduling system. In: Job scheduling strategies for parallel processing. Springer Berlin, Heidelberg, pp 295–303 (1995)

  27. Mu’alem AW, Feitelson DG (2001) Utilization, predictability, workloads, and user runtime estimates in scheduling the IBM SP2 with backfilling. IEEE Trans Parallel Distrib Syst 12(6):529–543

    Article  Google Scholar 

  28. MPI library (2013) http://www.mcs.anl.gov/research/projects/mpi/. Accessed 7 Mar 2014

  29. MacDougall MH (1987) Simulating computer systems: techniques and tools. MIT Press, Cambridge

    Google Scholar 

  30. Subhlok J, Venkataramaiah S, Singh A (2002) Characterizing NAS benchmark performance on shared heterogeneous networks. In: Proceedings of the 16th international parallel and distributed processing symposium, IPDPS ’02. IEEE Computer Society, Washington, pp 91

  31. Intel MPI Benchmarks (2011) http://software.intel.com/en-us/articles/intel-mpi-benchmarks. Accessed 7 Mar 2014

Download references

Acknowledgments

This work was supported by the Ministry of Science and Technology of Spain under contract TIN2012-34557

Author information

Authors and Affiliations

  1. Computer Architecture Department, Universitat Politècnica de Catalunya. BarcelonaTech, Jordi Girona 1-3, Campus Nord, D6-102, 08034 , Barcelona, Spain

    Gladys Utrera

  2. Computer Architecture Department, Universitat Politècnica de Catalunya. BarcelonaTech, Jordi Girona 1-3, Campus Nord, C6-203, 08034 , Barcelona, Spain

    Julita Corbalan

  3. Barcelona Supercomputing Center, Jordi Girona 1-3, Campus Nord, K2M-Planta 3, 08034 , Barcelona, Spain

    Jesús Labarta

Authors
  1. Gladys Utrera
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Julita Corbalan
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jesús Labarta
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Gladys Utrera.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Utrera, G., Corbalan, J. & Labarta, J. Scheduling parallel jobs on multicore clusters using CPU oversubscription. J Supercomput 68, 1113–1140 (2014). https://doi.org/10.1007/s11227-014-1142-9

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11227-014-1142-9

Keywords

Search

Navigation