TH-MPI: OS Kernel Integrated Fault Tolerant MPI

Consisting of large numbers of computing nodes, parallel cluster systems have high risks of individual node failure. To overcome the high overhead drawbacks of current fault tolerant MPI systems, this paper presents TH-MPI for parallel cluster systems. Being integrated into Linux kernel, THMPI is implemented in a more effective, transparent and extensive way. With supports of dynamic kernel module and diskless checkpointing technologies, our experiment shows that checkpointing in TH-MPI is effectively optimized.

[1]  Roy Friedman,et al.  Starfish: Fault-Tolerant Dynamic MPI Programs on Clusters of Workstations , 1999, Proceedings. The Eighth International Symposium on High Performance Distributed Computing (Cat. No.99TH8469).

[2]  Georg Stellner,et al.  CoCheck: checkpointing and process migration for MPI , 1996, Proceedings of International Conference on Parallel Processing.

[3]  Miron Livny,et al.  Condor-a hunter of idle workstations , 1988, [1988] Proceedings. The 8th International Conference on Distributed.

[4]  Kai Li,et al.  Libckpt: Transparent Checkpointing under UNIX , 1995, USENIX.