Interpreting write performance of supercomputer I/O systems with regression models

Bing Xie; Zilong Tan; Philip Carns; Jeff Chase; Kevin Harms; Jay Lofstead; Sarp Oral; Sudharshan S. Vazhkudai; Feiyi Wang

doi:10.1109/IPDPS49936.2021.00064

Interpreting write performance of supercomputer I/O systems with regression models

Bing Xie, Zilong Tan, Philip Carns, Jeff Chase, Kevin Harms, Jay Lofstead, Sarp Oral, Sudharshan S. Vazhkudai, Feiyi Wang

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

9 Scopus citations

Abstract

This work seeks to advance the state of the art in HPC I/O performance analysis and interpretation. In particular, we demonstrate effective techniques to: (1) model output performance in the presence of I/O interference from production loads; (2) build features from write patterns and key parameters of the system architecture and configurations; (3) employ suitable machine learning algorithms to improve model accuracy. We train models with five popular regression algorithms and conduct experiments on two distinct production HPC platforms. We find that the lasso and random forest models predict output performance with high accuracy on both of the target systems. We also explore use of the models to guide adaptation in I/O middleware systems, and show potential for improvements of at least 15% from model-guided adaptation on 70% of samples, and improvements up to 10 × on some samples for both of the target systems.

Original language	English
Title of host publication	Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	557-566
Number of pages	10
ISBN (Electronic)	9781665440660
DOIs	https://doi.org/10.1109/IPDPS49936.2021.00064
State	Published - May 2021
Event	35th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2021 - Virtual, Online Duration: May 17 2021 → May 21 2021

Publication series

Name	Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021

Conference

Conference	35th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2021
City	Virtual, Online
Period	05/17/21 → 05/21/21

Funding

ACKNOWLEDGMENTS This research used resources of the Oak Ridge Leadership Computing Facility, located in the National Center for Computational Sciences at the Oak Ridge National Laboratory, which is supported by the Office of Science of the Department of Energy under Contract DE-AC05-00OR22725. This work was supported by the U.S. Department of Energy, Office of Science, Advanced Scientific Computing Research, under Contract DE-AC02-06CH11357. This research used resources of the Argonne Leadership Computing Facility at Argonne National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under contract DE-AC02-06CH11357. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Keywords

High performance computing
I/O performance
Machine learning

Access to Document

10.1109/IPDPS49936.2021.00064

Cite this

Xie, B., Tan, Z., Carns, P., Chase, J., Harms, K., Lofstead, J., Oral, S., Vazhkudai, S. S., & Wang, F. (2021). Interpreting write performance of supercomputer I/O systems with regression models. In Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021 (pp. 557-566). Article 9460466 (Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IPDPS49936.2021.00064

Xie, Bing ; Tan, Zilong ; Carns, Philip et al. / Interpreting write performance of supercomputer I/O systems with regression models. Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 557-566 (Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021).

@inproceedings{752dd75c679e4cb2bca626e314695aca,

title = "Interpreting write performance of supercomputer I/O systems with regression models",

abstract = "This work seeks to advance the state of the art in HPC I/O performance analysis and interpretation. In particular, we demonstrate effective techniques to: (1) model output performance in the presence of I/O interference from production loads; (2) build features from write patterns and key parameters of the system architecture and configurations; (3) employ suitable machine learning algorithms to improve model accuracy. We train models with five popular regression algorithms and conduct experiments on two distinct production HPC platforms. We find that the lasso and random forest models predict output performance with high accuracy on both of the target systems. We also explore use of the models to guide adaptation in I/O middleware systems, and show potential for improvements of at least 15% from model-guided adaptation on 70% of samples, and improvements up to 10 × on some samples for both of the target systems.",

keywords = "High performance computing, I/O performance, Machine learning",

author = "Bing Xie and Zilong Tan and Philip Carns and Jeff Chase and Kevin Harms and Jay Lofstead and Sarp Oral and Vazhkudai, {Sudharshan S.} and Feiyi Wang",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 35th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2021 ; Conference date: 17-05-2021 Through 21-05-2021",

year = "2021",

month = may,

doi = "10.1109/IPDPS49936.2021.00064",

language = "English",

series = "Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021",

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

pages = "557--566",

booktitle = "Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021",

}

Xie, B, Tan, Z, Carns, P, Chase, J, Harms, K, Lofstead, J, Oral, S, Vazhkudai, SS & Wang, F 2021, Interpreting write performance of supercomputer I/O systems with regression models. in Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021., 9460466, Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021, Institute of Electrical and Electronics Engineers Inc., pp. 557-566, 35th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2021, Virtual, Online, 05/17/21. https://doi.org/10.1109/IPDPS49936.2021.00064

Interpreting write performance of supercomputer I/O systems with regression models. / Xie, Bing; Tan, Zilong; Carns, Philip et al.
Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021. Institute of Electrical and Electronics Engineers Inc., 2021. p. 557-566 9460466 (Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021).

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

TY - GEN

T1 - Interpreting write performance of supercomputer I/O systems with regression models

AU - Xie, Bing

AU - Tan, Zilong

AU - Carns, Philip

AU - Chase, Jeff

AU - Harms, Kevin

AU - Lofstead, Jay

AU - Oral, Sarp

AU - Vazhkudai, Sudharshan S.

AU - Wang, Feiyi

PY - 2021/5

Y1 - 2021/5

N2 - This work seeks to advance the state of the art in HPC I/O performance analysis and interpretation. In particular, we demonstrate effective techniques to: (1) model output performance in the presence of I/O interference from production loads; (2) build features from write patterns and key parameters of the system architecture and configurations; (3) employ suitable machine learning algorithms to improve model accuracy. We train models with five popular regression algorithms and conduct experiments on two distinct production HPC platforms. We find that the lasso and random forest models predict output performance with high accuracy on both of the target systems. We also explore use of the models to guide adaptation in I/O middleware systems, and show potential for improvements of at least 15% from model-guided adaptation on 70% of samples, and improvements up to 10 × on some samples for both of the target systems.

AB - This work seeks to advance the state of the art in HPC I/O performance analysis and interpretation. In particular, we demonstrate effective techniques to: (1) model output performance in the presence of I/O interference from production loads; (2) build features from write patterns and key parameters of the system architecture and configurations; (3) employ suitable machine learning algorithms to improve model accuracy. We train models with five popular regression algorithms and conduct experiments on two distinct production HPC platforms. We find that the lasso and random forest models predict output performance with high accuracy on both of the target systems. We also explore use of the models to guide adaptation in I/O middleware systems, and show potential for improvements of at least 15% from model-guided adaptation on 70% of samples, and improvements up to 10 × on some samples for both of the target systems.

KW - High performance computing

KW - I/O performance

KW - Machine learning

UR - http://www.scopus.com/inward/record.url?scp=85113557490&partnerID=8YFLogxK

U2 - 10.1109/IPDPS49936.2021.00064

DO - 10.1109/IPDPS49936.2021.00064

M3 - Conference contribution

AN - SCOPUS:85113557490

T3 - Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021

SP - 557

EP - 566

BT - Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 35th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2021

Y2 - 17 May 2021 through 21 May 2021

ER -

Xie B, Tan Z, Carns P, Chase J, Harms K, Lofstead J et al. Interpreting write performance of supercomputer I/O systems with regression models. In Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021. Institute of Electrical and Electronics Engineers Inc. 2021. p. 557-566. 9460466. (Proceedings - 2021 IEEE 35th International Parallel and Distributed Processing Symposium, IPDPS 2021). doi: 10.1109/IPDPS49936.2021.00064

Interpreting write performance of supercomputer I/O systems with regression models

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this