loading...
 This Article 
   
 Share 
   
 Bibliographic References 
   
 Add to: 
 
Digg
Furl
Spurl
Blink
Simpy
Google
Del.icio.us
Y!MyWeb
 
 Search 
   
19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Workshop 1
Event Logging: Portable and Efficient Checkpointing in Heterogeneous Environments with Non-FIFO Communication Platforms
Denver, Colorado
April 04-April 08
ISBN: 0-7695-2312-9
Zhao Peng, University College Dublin, Belfield, Ireland
Alexey Lastovetsky, University College Dublin, Belfield, Ireland
The Chandy-Lamport checkpointing algorithm is widely used in fault tolerant implementations of MPI. However, it assumes the FIFO property of message passing, which is not guaranteed by the MPI standard at the application level. Therefore, this algorithm cannot serve as a basis for an implementation-independent fault tolerant MPI.
In this paper, we present a variant of the Chandy-Lamport algorithm that does not rely on the FIFO property. This algorithm can be implemented on top of MPI and, hence, used for development of a supplement software component enabling the fault tolerance of any MPI implementation compliant with the MPI standard. We prove the correctness of the algorithm and analyze its performance. Experimental results demonstrating the efficiency of the algorithm are also presented.
Citation:
Zhao Peng, Alexey Lastovetsky, "Event Logging: Portable and Efficient Checkpointing in Heterogeneous Environments with Non-FIFO Communication Platforms," ipdps, vol. 2, pp.125b, 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Workshop 1, 2005
Usage of this product signifies your acceptance of the Terms of Use.