Advertisement

Experiments with Migration of Message-Passing Tasks

  • K. A. Iskra
  • Z. W. Hendrikse
  • G. D. van Albada
  • B. J. Overeinder
  • P. M. A. Sloot
  • J. Gehring
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1971)

Abstract

The combined computing capacity of the workstations that are present in many organisations nowadays is often under-utilised, as the performance for parallel programs is unpredictable. Load balancing through dynamic task re-allocation can help to obtain a more reliable performance. The Esprit project Dynamite1 provides such an automated load balancing system. It can migrate tasks that are part of a parallel program using a message passing library. Currently Dynamite supports PVM only, but it is being extended to support MPI as well. The Dynamite package is completely transparent, i.e. neither system (kernel) nor application source code need to be modified. Dynamite supports migration of tasks using dynamically linked libraries, open files and both direct and indirect PVM communication. Monitors and a scheduler are included. In this paper, we first briefly describe the Dynamite system. Next we describe how migration decisions are made and report on some performance measurements.

Keywords

Parallel Program Parallel Application Hill Climbing Migration Decision Dynamic Load Balance 
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.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Albada, G. D.van, Clinckemaillie, J., Emmen, A. H. L., Gehring, J., Heinz, O., Linden, F. van der, Overeinder, B. J., Reinefeld, A., Sloot, P. M. A.: Dynamite-blasting obstacles to parallel cluster computing. in Sloot, P. M. A., Bubak, M., Hoekstra, A. G., Hertzberger, L. O., editors, High-Performance Computing and Networking (HPCN Europe’ 99), LNCS 1593 300–310CrossRefGoogle Scholar
  2. 2.
    Casas, J., Clark, D. L., Konuru, R., Otto, S. W., Prouty, R. M., Walpole, J.: MPVM A Migration Transparant Version of PVM. Computer Systems 8 nr 2 (1995) 171–216Google Scholar
  3. 3.
    Czarnul, P., Krawczyk, H.: Dynamic Allocation with Process Migration in Distributed Environments. in Dongarra, J. J., Luque, E., Margalef, T., editors, Recent Advances in Parallel Virtual Machine and Message Passing Interface: 6th European PVM/MPI Users’ Group Meeting, LNCS 1697 (1999) 509–516CrossRefGoogle Scholar
  4. 4.
    Dan, P., Dongsheng, W., Youhui, Z., Meiming, S.: Quasi-asynchronous Migration: A Novel Migration Protocol for PVM Tasks. Operating Systems Review 33 nr 2 (1999) 5–14CrossRefGoogle Scholar
  5. 5.
    Geist, A., Beguelin, A., Dongarra, J., Jiang, W., Mancheck, R., Sunderam, V.: PVM: Parallel Virtual Machine. A Users’ Guide and Tutorial for Networked Parallel Computing. MIT Press, Cambridge, Massachusetts (1994) http://www.epm.ornl.gov/pvm/
  6. 6.
    Iskra, K. A., Hendrikse, Z. W., Albada, G. D. van, Overeinder, B. J., Sloot, P. M. A.: Experiments with Migration of PVM Tasks. in ISThmus 2000, Research and Development for the Information Society, Conference Proceedings, Poznan, Poland (2000) 295–304Google Scholar
  7. 7.
    Iskra, K. A., Hendrikse, Z. W., Albada, G. D. van, Overeinder, B. J., Sloot, P. M. A.: Performance Measurements on Dynamite/DPVM. in Dongarra, J., Kacsuk, P., Podhorszki, N., editors, Recent Advances in PVM and MPI. 7th European PVM/MPI User’s Group Meeting, LNCS 1908 (2000) (in press)Google Scholar
  8. 8.
    Iskra, K. A., Linden, F. van der, Hendrikse, Z. W., Albada, G. D. van, Overeinder, B. J., Sloot, P. M. A.: The implementation of Dynamite-an environment for migrating PVM tasks. Operating Systems Review nr 3 (2000) 40–55Google Scholar
  9. 9.
    Litzkow, M., Tannenbaum, T., Basney, J., Livny, M.: Checkpoint and migration of Unix processes in the Condor distributed processing system. Technical Report 1346, University of Wisconsin, WI, USA (1997)Google Scholar
  10. 10.
    Livny, M., Pruyne, J.: Managing Checkpoints for Parallel Programs, in Rudolph, L., Feitelson, D. G. editors: Proceedings IPPS Second Workshop on Job Scheduling Strategies for Parallel Processing, LNCS 1162 (1996) 140–154Google Scholar
  11. 11.
    Overeinder, B. J., Sloot, P. M. A., Heederik, R. N., Hertzberger, L. O.: A dynamic load balancing system for parallel cluster computing. Future Generation Computer Systems 12 (1996) 101–115CrossRefGoogle Scholar
  12. 12.
    Robinson, J., Russ, S. H., Flachs, B., Heckel, B.: A task migration implementation of the Message Passing Interface. Proceedings of the 5th IEEE international symposium on high performance distributed computing (1996) 61–68Google Scholar
  13. 13.
    Ronde, J. F. de, Albada, G. D. van, Sloot, P. M. A.: High Performance Simulation of Gravitational Radiation Antennas, in L. O. Hertzberger, P. M. A. Sloot, editors, High Performance Computing and Networking, LNCS 1225 (1997) 200–212CrossRefGoogle Scholar
  14. 14.
    Ronde, J. F. de, Albada, G. D. van, Sloot, P. M. A.: Simulation of GravitationalWave Detectors. Computers in Physics, 11 nr 5 (1997) 484–497CrossRefGoogle Scholar
  15. 15.
    Tan, C. P., Wong, W. F., Yuen, C. K.: Task Migratable PVM. Proceedings of the 2nd Merged Symposium IPPS/SPDP. (1999) 196–202Google Scholar
  16. 16.
    MPI: A Message-Passing Interface Standard, Version 1.1. Technical Report, University of Tennessee, Knoxville (1995) http://www-unix.mcs.anl.gov/mpi/

Copyright information

© Springer-Verlag Berlin Heidelberg 2000

Authors and Affiliations

  • K. A. Iskra
    • 1
  • Z. W. Hendrikse
    • 1
  • G. D. van Albada
    • 1
  • B. J. Overeinder
    • 1
  • P. M. A. Sloot
    • 1
  • J. Gehring
    • 2
  1. 1.Informatics InstituteUniversiteit van AmsterdamAmsterdamThe Netherlands
  2. 2.Paderborn Center for Parallel ComputingPaderbornGermany

Personalised recommendations