Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications

Bo Fang; Jieyang Chen; Karthik Pattabiraman; Matei Ripeanu; Sriram Krishnamoorthy

doi:10.1109/DSN-S.2019.00012

Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications

Bo Fang
, Jieyang Chen
, Karthik Pattabiraman
, Matei Ripeanu
, Sriram Krishnamoorthy

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

The roll-forward recovery schemes on HPC systems implicitly trade off faster time to solution for higher risk: as it usually performs a probabilistic repair, this may cause further failures such as SDCs. It is essential for users to be able to reason about the impact of a particular repair exercised by the scheme. Towards this goal, we identify two research questions aiming to determine the outcome of a repair either at the failure point or at the end of the execution. For the former, we propose a promising hybrid approach that combines machine learning and error propagation analysis techniques.

Original language	English
Title of host publication	Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	13-14
Number of pages	2
ISBN (Electronic)	9781728130286
DOIs	https://doi.org/10.1109/DSN-S.2019.00012
State	Published - Jun 2019
Externally published	Yes
Event	49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN-S 2019 - Portland, United States Duration: Jun 24 2019 → Jun 27 2019

Publication series

Name	Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019

Conference

Conference	49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN-S 2019
Country/Territory	United States
City	Portland
Period	06/24/19 → 06/27/19

Funding

ACKNOWLEDGEMENT This work was supported in part by the U.S. Department of Energy‘s (DOE) Office of Science, the National Sciences and Engineering Research Council of Canada (NSERC), Office of AdvancedScientific Computing Research, under award 66905. Pacific Northwest National Laboratory is operated by Battelle for DOE under Contract DE-AC05-76RL01830. This work was supported in part by the U.S. Department of Energy's (DOE) Office of Science, the National Sciences and Engineering Research Council of Canada (NSERC), Office of Advanced Scientific Computing Research, under award 66905. Pacific Northwest National Laboratory is operated by Battelle for DOE under Contract DE-AC05-76RL01830.

Keywords

Checkpoint/restart
Fault tolerance
HPC
Machine learning
Roll forward recovery

Access to Document

10.1109/DSN-S.2019.00012

Cite this

Fang, B., Chen, J., Pattabiraman, K., Ripeanu, M., & Krishnamoorthy, S. (2019). Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications. In Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019 (pp. 13-14). Article 8805765 (Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/DSN-S.2019.00012

Fang, Bo ; Chen, Jieyang ; Pattabiraman, Karthik et al. / Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications. Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 13-14 (Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019).

@inproceedings{46015e9d10ae4bb99fbaf01492395b9c,

title = "Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications",

abstract = "The roll-forward recovery schemes on HPC systems implicitly trade off faster time to solution for higher risk: as it usually performs a probabilistic repair, this may cause further failures such as SDCs. It is essential for users to be able to reason about the impact of a particular repair exercised by the scheme. Towards this goal, we identify two research questions aiming to determine the outcome of a repair either at the failure point or at the end of the execution. For the former, we propose a promising hybrid approach that combines machine learning and error propagation analysis techniques.",

keywords = "Checkpoint/restart, Fault tolerance, HPC, Machine learning, Roll forward recovery",

author = "Bo Fang and Jieyang Chen and Karthik Pattabiraman and Matei Ripeanu and Sriram Krishnamoorthy",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.; 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN-S 2019 ; Conference date: 24-06-2019 Through 27-06-2019",

year = "2019",

month = jun,

doi = "10.1109/DSN-S.2019.00012",

language = "English",

series = "Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "13--14",

booktitle = "Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019",

}

Fang, B, Chen, J, Pattabiraman, K, Ripeanu, M & Krishnamoorthy, S 2019, Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications. in Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019., 8805765, Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019, Institute of Electrical and Electronics Engineers Inc., pp. 13-14, 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN-S 2019, Portland, United States, 06/24/19. https://doi.org/10.1109/DSN-S.2019.00012

Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications. / Fang, Bo; Chen, Jieyang; Pattabiraman, Karthik et al.
Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 13-14 8805765 (Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications

AU - Fang, Bo

AU - Chen, Jieyang

AU - Pattabiraman, Karthik

AU - Ripeanu, Matei

AU - Krishnamoorthy, Sriram

PY - 2019/6

Y1 - 2019/6

N2 - The roll-forward recovery schemes on HPC systems implicitly trade off faster time to solution for higher risk: as it usually performs a probabilistic repair, this may cause further failures such as SDCs. It is essential for users to be able to reason about the impact of a particular repair exercised by the scheme. Towards this goal, we identify two research questions aiming to determine the outcome of a repair either at the failure point or at the end of the execution. For the former, we propose a promising hybrid approach that combines machine learning and error propagation analysis techniques.

AB - The roll-forward recovery schemes on HPC systems implicitly trade off faster time to solution for higher risk: as it usually performs a probabilistic repair, this may cause further failures such as SDCs. It is essential for users to be able to reason about the impact of a particular repair exercised by the scheme. Towards this goal, we identify two research questions aiming to determine the outcome of a repair either at the failure point or at the end of the execution. For the former, we propose a promising hybrid approach that combines machine learning and error propagation analysis techniques.

KW - Checkpoint/restart

KW - Fault tolerance

KW - HPC

KW - Machine learning

KW - Roll forward recovery

UR - https://www.scopus.com/pages/publications/85072010732

U2 - 10.1109/DSN-S.2019.00012

DO - 10.1109/DSN-S.2019.00012

M3 - Conference contribution

AN - SCOPUS:85072010732

T3 - Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019

SP - 13

EP - 14

BT - Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, DSN-S 2019

Y2 - 24 June 2019 through 27 June 2019

ER -

Fang B, Chen J, Pattabiraman K, Ripeanu M, Krishnamoorthy S. Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications. In Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 13-14. 8805765. (Proceedings - 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Supplemental Volume, DSN-S 2019). doi: 10.1109/DSN-S.2019.00012

Towards Predicting the Impact of Roll-Forward Failure Recovery for HPC Applications

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this