Compile-time analysis of communicating processes
ICS '92 Proceedings of the 6th international conference on Supercomputing
Efficient checkpointing on MIMD architectures
Efficient checkpointing on MIMD architectures
Time, clocks, and the ordering of events in a distributed system
Communications of the ACM
Broadway: A Software Architecture for Scientific Computing
Proceedings of the IFIP TC2/WG2.5 Working Conference on the Architecture of Scientific Software
Automated application-level checkpointing of MPI programs
Proceedings of the ninth ACM SIGPLAN symposium on Principles and practice of parallel programming
Compiler Support for Automatic Checkpointing
HPCS '02 Proceedings of the 16th Annual International Symposium on High Performance Computing Systems and Applications
SPDP '96 Proceedings of the 8th IEEE Symposium on Parallel and Distributed Processing (SPDP '96)
Consistent Global States of Distributed Systems: Fundamental Concepts and Mechanisms
Consistent Global States of Distributed Systems: Fundamental Concepts and Mechanisms
Compiler-Assisted Checkpointing
Compiler-Assisted Checkpointing
Cooperative checkpointing: a robust approach to large-scale systems reliability
Proceedings of the 20th annual international conference on Supercomputing
TH-1: China's first petaflop supercomputer
Frontiers of Computer Science in China
Hi-index | 0.00 |
To minimize work lost due to system failures, large parallel applications perform periodic checkpoints. These checkpoints are typically inserted manually by application programmers, resulting in synchronous checkpoints, or checkpoints that occur at the same program point in all processes. While this solution is tenable for current systems, it will become problematic for future supercomputers that have many tens of thousands of nodes, because contention for both the network and file system will grow. This paper shows that staggered checkpoints---globally consistent checkpoints in which processes perform checkpoints at different points in the code---can significantly reduce network and file system contention. We describe a compiler-based approach for inserting staggered checkpoints, and we show, using trace-driven simulation, that staggered checkpointing is 23 times faster that synchronous checkpointing.