Experiments with Migration of Message-Passing Tasks

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 Dynamite 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.

[1]  Jack Dongarra,et al.  Proceedings of the 13th European PVM/MPI User's Group conference on Recent advances in parallel virtual machine and message passing interface , 1997 .

[2]  Peter M. A. Sloot,et al.  A dynamic load balancing system for parallel cluster computing , 1996, Future Gener. Comput. Syst..

[3]  Miron Livny,et al.  Checkpoint and Migration of UNIX Processes in the Condor Distributed Processing System , 1997 .

[4]  Wang Dongsheng,et al.  Quasi-asynchronous migration: a novel migration protocol for PVM tasks , 1999 .

[5]  Miron Livny,et al.  Managing Checkpoints for Parallel Programs , 1996, JSSPP.

[6]  Peter M. A. Sloot,et al.  The implementation of dynamite: an environment for migrating PVM tasks , 2000, OPSR.

[7]  Peter M. A. Sloot,et al.  Performance Measurements on Dynamite/DPVM , 2000, PVM/MPI.

[8]  Nigel P. Topham,et al.  Performance of the decoupled ACRI-1 architecture: the perfect club , 1995, HPCN Europe.

[9]  G. D. van Albada,et al.  High performance simulation of gravitational radiation antennas , 1997 .

[10]  Peter M. A. Sloot,et al.  Experiments with Migration of PVM Tasks , 2000 .

[11]  Weng-Fai Wong,et al.  tmPVM-task migratable PVM , 1999, Proceedings 13th International Parallel Processing Symposium and 10th Symposium on Parallel and Distributed Processing. IPPS/SPDP 1999.

[12]  Peter M. A. Sloot,et al.  Dynamite - Blasting Obstacles to Parallel Cluster Computing , 1999, HPCN Europe.

[13]  Henryk Krawczyk,et al.  Dynamic Assignment with Process Migration in Distributed Environments , 1999, PVM/MPI.

[14]  Jonathan Robinson,et al.  A task migration implementation of the Message-Passing Interface , 1996, Proceedings of 5th IEEE International Symposium on High Performance Distributed Computing.

[15]  Peter M. A. Sloot,et al.  High Performance Simulation for Resonant-Mass Gravitational Radiation Antennas , 1997, HPCN Europe.

[16]  Jonathan Walpole,et al.  MPVM: A Migration Transparent Version of PVM , 1995, Comput. Syst..

[17]  Jack Dongarra,et al.  PVM: Parallel virtual machine: a users' guide and tutorial for networked parallel computing , 1995 .

[18]  Peter M. A. Sloot,et al.  Simulation of gravitational wave detectors , 1998 .