The overhead model of word-level and page-level incremental checkpointing

Authors:
Junyoung Heo;Yookun Cho;Gwangil Jeon;Haklin Kimm
Affiliations:
Seoul National University;Seoul National University;Korea Polytechnic University;East Stroudsburg University of Pennsylvania
Venue:
Proceedings of the 2006 ACM symposium on Applied computing
Year:
2006

Citing 4
Cited 0

Memory exclusion: optimizing the performance of checkpointing systems

Software—Practice & Experience
Efficient Incremental Checkpointing of Java Programs

DSN '00 Proceedings of the 2000 International Conference on Dependable Systems and Networks (formerly FTCS-30 and DCCA-8)
Adaptive incremental checkpointing for massively parallel systems

Proceedings of the 18th annual international conference on Supercomputing
Libckpt: transparent checkpointing under Unix

TCON'95 Proceedings of the USENIX 1995 Technical Conference Proceedings

Quantified Score

Hi-index	0.00

Visualization

Abstract

Generally, word-level granularity for incremental checkpointing may reduce the checkpoint file size, and hence get better the performance. However, word-level granularity may not always be more efficient than page-level granularity, because word-level granularity may sometimes increase the checkpointing overhead for finding the differences between two checkpoints and for writing the address of the modified word to stable storage. In this paper, we make the overhead model and find factors which produce an effect on the over-head. We also show that the model and the factors are fairly reasonable via experimental results from Linux Kernel-level incremental checkpointing facility.