Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables

Qiao Kang; Scot Breitenfeld; Kaiyuan Hou; Wei Keng Liao; Robert Ross; Suren Byna

doi:10.1109/BigData52589.2021.9671638

Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables

Qiao Kang, Scot Breitenfeld, Kaiyuan Hou, Wei Keng Liao, Robert Ross, Suren Byna

Electrical and Computer Engineering

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

3 Scopus citations

Abstract

Accessing non-contiguous blocks in multiple array variables is a challenging I/O pattern for parallel applications to obtain good I/O performance. High-level I/O libraries such as HDF5 allow users to implement this pattern conveniently, but users have observed significant performance bottlenecks in the two-phase I/O implementation of MPI-IO. Recent studies have advanced the two-phase I/O performance by novel communication algorithms, but such improvements still have limitations. Two-phase I/O has to faithfully process inputs from high-level I/O libraries, so that implementation overheads can accumulate for improper usage of high-level I/O libraries. In this paper, we propose approaches for efficient usage of high-level I/O libraries that can circumvent major collective I/O overheads. We adopt a multi-dataset implementation of HDF5 dataset I/O to aggregate non-contiguous requests for array blocks and provide corresponding parameter assignment strategies. These approaches reduce the overheads caused by communication straggler effects in two-phase I/O. We show that our proposed methods can improve the parallel I/O performance up to 8× on two supercomputing systems for the HDF5 implementations of an I/O kernel extracted from climate simulation code compared with its baseline implementations.

Original language	English (US)
Title of host publication	Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021
Editors	Yixin Chen, Heiko Ludwig, Yicheng Tu, Usama Fayyad, Xingquan Zhu, Xiaohua Tony Hu, Suren Byna, Xiong Liu, Jianping Zhang, Shirui Pan, Vagelis Papalexakis, Jianwu Wang, Alfredo Cuzzocrea, Carlos Ordonez
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	98-108
Number of pages	11
ISBN (Electronic)	9781665439022
DOIs	https://doi.org/10.1109/BigData52589.2021.9671638
State	Published - 2021
Event	2021 IEEE International Conference on Big Data, Big Data 2021 - Virtual, Online, United States Duration: Dec 15 2021 → Dec 18 2021

Publication series

Name	Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021

Conference

Conference	2021 IEEE International Conference on Big Data, Big Data 2021
Country/Territory	United States
City	Virtual, Online
Period	12/15/21 → 12/18/21

Funding

This research was supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration. This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. This research used resources of the Oak Ridge Leadership Computing Facility at the Oak Ridge National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. 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.

ASJC Scopus subject areas

Information Systems and Management
Artificial Intelligence
Computer Vision and Pattern Recognition
Information Systems

Access to Document

10.1109/BigData52589.2021.9671638

Cite this

Kang, Q., Breitenfeld, S., Hou, K., Liao, W. K., Ross, R., & Byna, S. (2021). Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables. In Y. Chen, H. Ludwig, Y. Tu, U. Fayyad, X. Zhu, X. T. Hu, S. Byna, X. Liu, J. Zhang, S. Pan, V. Papalexakis, J. Wang, A. Cuzzocrea, & C. Ordonez (Eds.), Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021 (pp. 98-108). (Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/BigData52589.2021.9671638

Kang, Qiao ; Breitenfeld, Scot ; Hou, Kaiyuan et al. / Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables. Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021. editor / Yixin Chen ; Heiko Ludwig ; Yicheng Tu ; Usama Fayyad ; Xingquan Zhu ; Xiaohua Tony Hu ; Suren Byna ; Xiong Liu ; Jianping Zhang ; Shirui Pan ; Vagelis Papalexakis ; Jianwu Wang ; Alfredo Cuzzocrea ; Carlos Ordonez. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 98-108 (Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021).

@inproceedings{9f9bb9747f14470da7f5d5e409bef9c4,

title = "Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables",

abstract = "Accessing non-contiguous blocks in multiple array variables is a challenging I/O pattern for parallel applications to obtain good I/O performance. High-level I/O libraries such as HDF5 allow users to implement this pattern conveniently, but users have observed significant performance bottlenecks in the two-phase I/O implementation of MPI-IO. Recent studies have advanced the two-phase I/O performance by novel communication algorithms, but such improvements still have limitations. Two-phase I/O has to faithfully process inputs from high-level I/O libraries, so that implementation overheads can accumulate for improper usage of high-level I/O libraries. In this paper, we propose approaches for efficient usage of high-level I/O libraries that can circumvent major collective I/O overheads. We adopt a multi-dataset implementation of HDF5 dataset I/O to aggregate non-contiguous requests for array blocks and provide corresponding parameter assignment strategies. These approaches reduce the overheads caused by communication straggler effects in two-phase I/O. We show that our proposed methods can improve the parallel I/O performance up to 8× on two supercomputing systems for the HDF5 implementations of an I/O kernel extracted from climate simulation code compared with its baseline implementations.",

author = "Qiao Kang and Scot Breitenfeld and Kaiyuan Hou and Liao, {Wei Keng} and Robert Ross and Suren Byna",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 2021 IEEE International Conference on Big Data, Big Data 2021 ; Conference date: 15-12-2021 Through 18-12-2021",

year = "2021",

doi = "10.1109/BigData52589.2021.9671638",

language = "English (US)",

series = "Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021",

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

pages = "98--108",

editor = "Yixin Chen and Heiko Ludwig and Yicheng Tu and Usama Fayyad and Xingquan Zhu and Hu, {Xiaohua Tony} and Suren Byna and Xiong Liu and Jianping Zhang and Shirui Pan and Vagelis Papalexakis and Jianwu Wang and Alfredo Cuzzocrea and Carlos Ordonez",

booktitle = "Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021",

address = "United States",

}

Kang, Q, Breitenfeld, S, Hou, K, Liao, WK, Ross, R & Byna, S 2021, Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables. in Y Chen, H Ludwig, Y Tu, U Fayyad, X Zhu, XT Hu, S Byna, X Liu, J Zhang, S Pan, V Papalexakis, J Wang, A Cuzzocrea & C Ordonez (eds), Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021. Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021, Institute of Electrical and Electronics Engineers Inc., pp. 98-108, 2021 IEEE International Conference on Big Data, Big Data 2021, Virtual, Online, United States, 12/15/21. https://doi.org/10.1109/BigData52589.2021.9671638

Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables. / Kang, Qiao; Breitenfeld, Scot; Hou, Kaiyuan et al.
Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021. ed. / Yixin Chen; Heiko Ludwig; Yicheng Tu; Usama Fayyad; Xingquan Zhu; Xiaohua Tony Hu; Suren Byna; Xiong Liu; Jianping Zhang; Shirui Pan; Vagelis Papalexakis; Jianwu Wang; Alfredo Cuzzocrea; Carlos Ordonez. Institute of Electrical and Electronics Engineers Inc., 2021. p. 98-108 (Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021).

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

TY - GEN

T1 - Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables

AU - Kang, Qiao

AU - Breitenfeld, Scot

AU - Hou, Kaiyuan

AU - Liao, Wei Keng

AU - Ross, Robert

AU - Byna, Suren

PY - 2021

Y1 - 2021

N2 - Accessing non-contiguous blocks in multiple array variables is a challenging I/O pattern for parallel applications to obtain good I/O performance. High-level I/O libraries such as HDF5 allow users to implement this pattern conveniently, but users have observed significant performance bottlenecks in the two-phase I/O implementation of MPI-IO. Recent studies have advanced the two-phase I/O performance by novel communication algorithms, but such improvements still have limitations. Two-phase I/O has to faithfully process inputs from high-level I/O libraries, so that implementation overheads can accumulate for improper usage of high-level I/O libraries. In this paper, we propose approaches for efficient usage of high-level I/O libraries that can circumvent major collective I/O overheads. We adopt a multi-dataset implementation of HDF5 dataset I/O to aggregate non-contiguous requests for array blocks and provide corresponding parameter assignment strategies. These approaches reduce the overheads caused by communication straggler effects in two-phase I/O. We show that our proposed methods can improve the parallel I/O performance up to 8× on two supercomputing systems for the HDF5 implementations of an I/O kernel extracted from climate simulation code compared with its baseline implementations.

AB - Accessing non-contiguous blocks in multiple array variables is a challenging I/O pattern for parallel applications to obtain good I/O performance. High-level I/O libraries such as HDF5 allow users to implement this pattern conveniently, but users have observed significant performance bottlenecks in the two-phase I/O implementation of MPI-IO. Recent studies have advanced the two-phase I/O performance by novel communication algorithms, but such improvements still have limitations. Two-phase I/O has to faithfully process inputs from high-level I/O libraries, so that implementation overheads can accumulate for improper usage of high-level I/O libraries. In this paper, we propose approaches for efficient usage of high-level I/O libraries that can circumvent major collective I/O overheads. We adopt a multi-dataset implementation of HDF5 dataset I/O to aggregate non-contiguous requests for array blocks and provide corresponding parameter assignment strategies. These approaches reduce the overheads caused by communication straggler effects in two-phase I/O. We show that our proposed methods can improve the parallel I/O performance up to 8× on two supercomputing systems for the HDF5 implementations of an I/O kernel extracted from climate simulation code compared with its baseline implementations.

UR - http://www.scopus.com/inward/record.url?scp=85125298495&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85125298495&partnerID=8YFLogxK

U2 - 10.1109/BigData52589.2021.9671638

DO - 10.1109/BigData52589.2021.9671638

M3 - Conference contribution

AN - SCOPUS:85125298495

T3 - Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021

SP - 98

EP - 108

BT - Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021

A2 - Chen, Yixin

A2 - Ludwig, Heiko

A2 - Tu, Yicheng

A2 - Fayyad, Usama

A2 - Zhu, Xingquan

A2 - Hu, Xiaohua Tony

A2 - Byna, Suren

A2 - Liu, Xiong

A2 - Zhang, Jianping

A2 - Pan, Shirui

A2 - Papalexakis, Vagelis

A2 - Wang, Jianwu

A2 - Cuzzocrea, Alfredo

A2 - Ordonez, Carlos

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2021 IEEE International Conference on Big Data, Big Data 2021

Y2 - 15 December 2021 through 18 December 2021

ER -

Kang Q, Breitenfeld S, Hou K, Liao WK, Ross R, Byna S. Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables. In Chen Y, Ludwig H, Tu Y, Fayyad U, Zhu X, Hu XT, Byna S, Liu X, Zhang J, Pan S, Papalexakis V, Wang J, Cuzzocrea A, Ordonez C, editors, Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021. Institute of Electrical and Electronics Engineers Inc. 2021. p. 98-108. (Proceedings - 2021 IEEE International Conference on Big Data, Big Data 2021). doi: 10.1109/BigData52589.2021.9671638

Optimizing Performance of Parallel I/O Accesses to Non-contiguous Blocks in Multiple Array Variables

Abstract

Publication series

Conference

Funding

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this