Characteristics of a Large Shared Memory Production Workload

  • Su-Hui Chiang
  • Mary K. Vernon
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2221)

Abstract

This paper characterizes the production workload that highly utilizes the NCSA Origin 2000. The characterization includes the distributions of job interarrival time, requested number of processors, requested memory, requested runtime, actual runtime as a fraction of requested runtime, and the ratio of memory usage to memory request. Conditional distributions are defined as needed for generating a synthetic workload with the same characteristics, including the key correlations observed among the job parameters. Characteristics of the O2K workload that differ from previously reported production workload characteristics are also noted.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Feitelson, D.G., Nitzberg, B.: Job characteristics of a production parallel scientific workload on the NASA Ames iPSC/860. In: Proc. 1stWorkshop on Job Scheduling Strategies for Parallel Processing, Santa Barbara, Lecture Notes in Comp. Sci. Vol. 949, Springer-Verlag (1995) 337–360Google Scholar
  2. 2.
    Hotovy, S.: Workload evolution on the Cornell Theory Center IBM SP2. In: Proc. 2nd Workshop on Job Scheduling Strategies for Parallel Processing, Honolulu, Lecture Notes in Comp. Sci. Vol. 1162, Springer-Verlag (1996) 27–40CrossRefGoogle Scholar
  3. 3.
    Feitelson, D.G.: Packing schemes for gang scheduling. In: Proc. 2nd Workshop on Job Scheduling Strategies for Parallel Processing, Honolulu, Lecture Notes in Comp. Sci. Vol. 1162, Springer-Verlag (1996) 89–110CrossRefGoogle Scholar
  4. 4.
    Subhlok, J., Gross, T., Suzuoka, T.: Impact of job mix on optimizations for space sharing schedulers. In: Proc. 1996 ACM/IEEE Supercomputing Conf., Pittsburgh (1996)Google Scholar
  5. 5.
    Hotovy, S., Schneider, D., O’Donnell, T.: Analysis of the early workload on the Cornell Theory Center IBM SP2. Technical Report TR234, Cornell Theory Center (1996)Google Scholar
  6. 6.
    Downey, A.B.: Predicting queue times on space-sharing parallel computers. In: Proc. 3rd Workshop on Job Scheduling Strategies for Parallel Processing, Geneva, Lecture Notes in Comp. Sci. Vol. 1291, Springer-Verlag (1997)Google Scholar
  7. 7.
    Feitelson, D.G.: Memory usage in the LANL CM-5 workload. In: Proc. 3rd Workshop on Job Scheduling Strategies for Parallel Processing, Geneva, Lecture Notes in Comp. Sci. Vol. 1291, Springer-Verlag (1997) 78–94Google Scholar
  8. 8.
    Windisch, K., Lo, V., Feitelson, D., Nitzberg, B., Moore, R.: A comparison of workload traces from two production parallel machines. In: Proc. 6th Symp. on the Frontiers of Massively Parallel Computation. (1996) 319–326Google Scholar
  9. 9.
    Jann, J., Pattnaik, P., Franke, H., Wang, F., Skovira, J., Riodan, J.: Modeling of workload in MPPs. In: Proc. 3rd Workshop on Job Scheduling Strategies for Parallel Processing, Geneva, Lecture Notes in Comp. Sci. Vol. 1291, Springer-Verlag (1997)Google Scholar
  10. 10.
    Squillante, M.S., Yao, D.D., Zhang, L.: The impact of job arrival patterns on parallel scheduling. Performance Evaluation Review 26 (1999) 52–59CrossRefGoogle Scholar
  11. 11.
    Setia, S.K., Squillante, M.S., Naik, V.K.: The impact of job memory requirements on gang-scheduling performance. Performance Evaluation Review 26 (1999) 30–39CrossRefGoogle Scholar
  12. 14.
    Chiang, S.H., Vernon, M.K.: Production job scheduling for parallel shared memory systems. In: Proc. Int’l. Parallel and Distributed Processing Symp. (IPDPS) 2001, San Francisco (2001)Google Scholar
  13. 15.
    Wan, M., Moore, R., Kremenek, G., Steube, K.: A batch scheduler for the Intel Paragon MPP system with a non-contiguous node allocation algorithm. In: Proc. 2nd Workshop on Job Scheduling Strategies for Parallel Processing, Honolulu, Lecture Notes in Comp. Sci. Vol. 1162, Springer-Verlag (1996) 48–64CrossRefGoogle Scholar
  14. 16.
    Downey, A.B., Feitelson, D.G.: The elusive goal of workload characterization. Performance Evaluation Review 26 (1999) 14–29CrossRefGoogle Scholar
  15. 17.
    Feitelson, D.G., Mu’alem Weil, A.: Utilization and predictability in scheduling the IBM SP2 with backfilling. In: Proc. 12th Int’l. Parallel Processing Symp., Orlando (1998) 542–546Google Scholar
  16. 18.
    Gibbons, R.: A historical application profiler for use by parallel schedulers. Master’s thesis, Univ. of Toronto, Ontario (1997)Google Scholar
  17. 19.
    Allen, A.O.: Probability, Statistics, and Queueing Theory with Computer Science Applications. 2nd edn. Academic Press (1990)Google Scholar
  18. 20.
    Johnson, N.L., Kotz, S., Kemp, A.W.: Univariate Discrete Distributions. 2nd edn. Wiley (1992)Google Scholar
  19. 22.
    Gibbons, R.: A historical application profiler for use by parallel schedulers. In: Proc. 3rd Workshop on Job Scheduling Strategies for Parallel Processing, Geneva, Lecture Notes in Comp. Sci. Vol. 1291, Springer-Verlag (1997)Google Scholar
  20. 23.
    Harchol-Balter, M., Downey, A.B.: Exploiting process lifetime distributions for dynamic load balancing. ACM Trans. on Computer Systems 15 (1997) 253–285CrossRefGoogle Scholar
  21. 24.
    Harchol-Balter, M.: Taskassignmen t with unknown duration. In: Proc. Int’l. Conf. on Distributed Computing Systems, Taipei, Taiwan (2000)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2001

Authors and Affiliations

  • Su-Hui Chiang
    • 1
  • Mary K. Vernon
    • 1
  1. 1.Computer Sciences DepartmentUniversity of WisconsinMadisonWisconsin

Personalised recommendations