Hiding checkpoint overhead in HPC applications with a semi-blocking algorithm

Xiang Ni, Esteban Meneses, Laxmikant V. Kalé

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The HPC community has seen a steady increase in the number of components in every generation of supercomputers. Assembling a large number of components into a single cluster makes a machine more powerful, but also much more prone to failures. Therefore, fault tolerance has become a major concern in HPC. To deal with node crashes in large systems, checkpoint/restart is by far the preferred method. A typical way to implement checkpoints is by using a blocking algorithm, which suspends the execution of the application while the checkpoint is safely stored. One limitation of the blocking algorithm is that it saturates the network bandwidth at the time of checkpoint. This problem will become even more critical because the projected network bandwidth increase will not match the increase in memory per node. To alleviate this problem, we have developed a semi-blocking checkpoint algorithm that overlaps execution of the application with transmission of checkpoints. Our implementation decomposes a checkpoint into small messages that are interleaved with application messages. The experimental results show a dramatic reduction in the checkpoint overhead for various applications. We present a model for our approach and use this model to compute the benefit of the semi-blocking algorithm for different failure rates predicted at Exascale. We estimate our method can reduce up to 22% the total execution time of an iterative scientific application.

Original languageEnglish (US)
Title of host publicationProceedings - 2012 IEEE International Conference on Cluster Computing, CLUSTER 2012
PublisherIEEE Computer Society
Pages364-372
Number of pages9
ISBN (Print)9780768548074
DOIs
StatePublished - Jan 1 2012
Event2012 IEEE International Conference on Cluster Computing, CLUSTER 2012 - Beijing, China
Duration: Sep 24 2012Sep 28 2012

Publication series

NameProceedings - 2012 IEEE International Conference on Cluster Computing, CLUSTER 2012

Other

Other2012 IEEE International Conference on Cluster Computing, CLUSTER 2012
CountryChina
CityBeijing
Period9/24/129/28/12

Keywords

  • SSD
  • adaptive runtime system
  • checkpoint/restart
  • fault tolerance
  • semi-blocking algorithm

ASJC Scopus subject areas

  • Software

Fingerprint Dive into the research topics of 'Hiding checkpoint overhead in HPC applications with a semi-blocking algorithm'. Together they form a unique fingerprint.

Cite this