Implications of I/O for gang scheduled workloads

  • Walter Lee
  • Matthew Frank
  • Victor Lee
  • Kenneth Mackenzie
  • Larry Rudolph
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1291)


The job workloads of general-purpose multiprocessors usually include both compute-bound parallel jobs, which often require gang scheduling, as well as I/O-bound jobs, which require high CPU priority for the individual gang members of the job in order to achieve interactive response times. Our results indicate that an effective interactive multiprocessor scheduler must be flexible and tailor the priority, time quantum, and extent of gang scheduling to the individual needs of each job.

Flexible gang scheduling is required because of several weaknesses of traditional gang scheduling. In particular, we show that the response time of I/O-bound jobs suffers under traditional gang scheduling. In addition, we show that not all applications benefit equally from gang scheduling; most real applications can tolerate at least a small amount of scheduling skew without major performance degradation. Finally, we show that messaging statistics contain information about whether applications require gang scheduling. Taken together these results provide evidence that flexible gang scheduling is both necessary and feasible.


Parallel Application Gang Member Load Imbalance Time Quantum Priority Inversion 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    R. Arpaci, A. Dusseau, A. Vahdat, L. Liu, T. Anderson, and D. Patterson. The Interaction of Parallel and Sequential Workloads on a Network of Workstations. In Proceedings of Sigmetrics/Performance '95, pages 267–278, May 1995.Google Scholar
  2. 2.
    J. M. Barton and N. Bitar. A Scalable Multi-Discipline, Multiple-Processor Scheduling Framework for IRIX. In Lecture Notes in Computer Science, 949, pages 45–69, Santa Barbara, 1995. Springer-Verlag. Workshop on Parallel Job Scheduling, IPPS '95.Google Scholar
  3. 3.
    M. Crovella, P. Das, C. Dubnicki, T. LeBlanc, and E. Markatos. Multiprogramming on Multiprocessors. In Proceedings of the third IEEE Symposium on Parallel and Distributed Processing, 1991.Google Scholar
  4. 4.
    R. Cypher, S. Konstantinidou, A. Ho, and P. Messina. A Quantitative Study of Parallel Scientific Applications with Explicit Communication. In The Journal of Supercomputing, pages 5–24, January 1996.Google Scholar
  5. 5.
    J. M. del Rosario and A. Choudhary. High Performance I/O for Parallel Computers: Problems and Prospects. In IEEE Computers, vol. 27, no. 3, pages 59–68, 1994.Google Scholar
  6. 6.
    A. Dusseau, R. Arpaci,, and D. Culler. Effective Distributed Scheduling of Parallel Workloads. In Proceedings of ACM SIGMETRICS 1996, Philadelphia, May 1996. ACM.Google Scholar
  7. 7.
    D. G. Feitelson. A Survey of Scheduling in Multiprogrammed Parallel Systems. Technical Report IBM/RC 19790(87657), IBM, October 1994.Google Scholar
  8. 8.
    D. G. Feitelson and L. Rudolph. Distributed Hierarchical Control for Parallel Processing. In Computer. IEEE, May 1990.Google Scholar
  9. 9.
    D. G. Feitelson and L. Rudolph. Gang Scheduling Performance Benefits for Fine-Grain Synchronization. In Journal of Parallel and Distributed Computing, pages 306–318, December 1992.Google Scholar
  10. 10.
    D. G. Feitelson and L. Rudolph. Coscheduling Based on Runtime Identification of Activity Working Sets. In International Journal of Parallel Programming, pages 135–160, April 1995.Google Scholar
  11. 11.
    A. Gupta, A. Tucker, and L. Stevens. The Impact of Operating System Scheduling Policies and Synchronization Methods of the Performance of Parallel Applications. In Proceedings of 1991 ACM Sigmetrics Conference, 1991.Google Scholar
  12. 12.
    K. L. Johnson, M. F. Kaashoek, and D. A. Wallach. CRL: High-Performance All Software Distributed Shared Memory. In Proceedings of the 15th ACM Symposium on Operating Systems Principles, December 1995.Google Scholar
  13. 13.
    K. Mackenzie, J. Kubiatowicz, M. Frank, W. Lee, V. Lee, A. Agarwal, and M. F. Kaashoek. UDM: User Direct Messaging for General-Purpose Multiprocessing. Technical Memo MIT/LCS/TM-556, March 1996.Google Scholar
  14. 14.
    R. Mraz. Reducing the Variance of Point-to-Point Transfers for Parallel Real-Time Programs. In IEEE Parallel & Distributed Technology, 1994.Google Scholar
  15. 15.
    J. K. Ousterhout. Scheduling Techniques for Concurrent Systems. In 3rd International Conference on Distributed Computing Systems, pages 22–30, 1982.Google Scholar
  16. 16.
    K. E. Schauser and C. J. Scheiman. Experience with Active Messages on the Meiko CS-2. In Proceedings of the 9th International Symposium on Parallel Processing, 1995.Google Scholar
  17. 17.
    P. G. Sobalvarro and W. E. Weihl. Demand-based Coscheduling of Parallel Jobs on Multiprogrammed Multiprocessors. In Lecture Notes in Computer Science, 949, pages 106–126, Santa Barbara, 1995. Springer Verlag. Workshop on Parallel Job Scheduling, IPPS '95.Google Scholar
  18. 18.
    A. Tucker and A. Gupta. Process Control and Scheduling Issues for Multiprogrammed Shared-Memory Multiprocessors. In Proceedings of the 12th ACM Symposium on Operating System Principles (SOSP-12), pages 159–166, December 1989.Google Scholar
  19. 19.
    F. Wang, H. Franke, M. Papaefthymiou, P. Pattnaik, L. Rudolph, and M. S. Squillante. A Gang Scheduling Design for Multiprogrammed Parallel Computing Environments. In Lecture Notes in Computer Science, 1162, pages 111–125, Honolulu, Hawaii, 1996. Springer Verlag. Workshop on Parallel Job Scheduling, IPPS '96.Google Scholar
  20. 20.
    K. Y. Wang and D. C. Marinescu. Correlation of the Paging Activity of the Individual Node Programs in the SPMD Execution Mode. In Proceedings of the Hawaii International Conference on System Sciences, 1995.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1997

Authors and Affiliations

  • Walter Lee
    • 1
  • Matthew Frank
    • 1
  • Victor Lee
    • 1
  • Kenneth Mackenzie
    • 1
  • Larry Rudolph
    • 1
  1. 1.M.I.T. Laboratory for Computer ScienceCambridgeUSA

Personalised recommendations