Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load

Jean Luca Bez; Ahmad Maroof Karimi; Arnab K. Paul; Bing Xie; Suren Byna; Philip Carns; Sarp Oral; Feiyi Wang; Jesse Hanley

doi:10.1145/3502181.3531461

Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load

Jean Luca Bez, Ahmad Maroof Karimi, Arnab K. Paul, Bing Xie, Suren Byna, Philip Carns, Sarp Oral, Feiyi Wang, Jesse Hanley

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

16 Scopus citations

Abstract

Scientific computing workloads at HPC facilities have been shifting from traditional numerical simulations to AI/ML applications for training and inference while processing and producing ever-increasing amounts of scientific data. To address the growing need for increased storage capacity, lower access latency, and higher bandwidth, emerging technologies such as non-volatile memory are integrated into supercomputer I/O subsystems. With these emerging trends, we need a better understanding of the multilayer supercomputer I/O systems and ways to use these subsystems efficiently. In this work, we study the I/O access patterns and performance characteristics of two representative supercomputer I/O subsystems. Through an extensive analysis of year-long I/O logs on each system, we report new observations in I/O reads and writes, unbalanced use of storage system layers, and new trends in user behaviors at the HPC I/O middleware stack.

Original language	English
Title of host publication	HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing
Publisher	Association for Computing Machinery, Inc
Pages	43-55
Number of pages	13
ISBN (Electronic)	9781450391993
DOIs	https://doi.org/10.1145/3502181.3531461
State	Published - Jun 27 2022
Event	31st International Symposium on High-Performance Parallel and Distributed Computing, HPDC 2022 - Virtual, Online, United States Duration: Jun 27 2022 → Jun 30 2022

Publication series

Name	HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing

Conference

Conference	31st International Symposium on High-Performance Parallel and Distributed Computing, HPDC 2022
Country/Territory	United States
City	Virtual, Online
Period	06/27/22 → 06/30/22

Funding

This work 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 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 under Contract No. DE-AC02-05CH11231. This work was supported by the U.S. Department of Energy, Office of Science, Advanced Scientific Computing Research, under Contract DE-AC02-06CH11357.

Keywords

access patterns
high-performance computing
in-system storage
parallel file systems
production system

Access to Document

10.1145/3502181.3531461

Cite this

Bez, J. L., Karimi, A. M., Paul, A. K., Xie, B., Byna, S., Carns, P., Oral, S., Wang, F., & Hanley, J. (2022). Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load. In HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing (pp. 43-55). (HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing). Association for Computing Machinery, Inc. https://doi.org/10.1145/3502181.3531461

Bez, Jean Luca ; Karimi, Ahmad Maroof ; Paul, Arnab K. et al. / Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load. HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing. Association for Computing Machinery, Inc, 2022. pp. 43-55 (HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing).

@inproceedings{10a52882ab6d47bb9ff4c01c73a73ecd,

title = "Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load",

abstract = "Scientific computing workloads at HPC facilities have been shifting from traditional numerical simulations to AI/ML applications for training and inference while processing and producing ever-increasing amounts of scientific data. To address the growing need for increased storage capacity, lower access latency, and higher bandwidth, emerging technologies such as non-volatile memory are integrated into supercomputer I/O subsystems. With these emerging trends, we need a better understanding of the multilayer supercomputer I/O systems and ways to use these subsystems efficiently. In this work, we study the I/O access patterns and performance characteristics of two representative supercomputer I/O subsystems. Through an extensive analysis of year-long I/O logs on each system, we report new observations in I/O reads and writes, unbalanced use of storage system layers, and new trends in user behaviors at the HPC I/O middleware stack.",

keywords = "access patterns, high-performance computing, in-system storage, parallel file systems, production system",

author = "Bez, {Jean Luca} and Karimi, {Ahmad Maroof} and Paul, {Arnab K.} and Bing Xie and Suren Byna and Philip Carns and Sarp Oral and Feiyi Wang and Jesse Hanley",

note = "Publisher Copyright: {\textcopyright} 2022 ACM.; 31st International Symposium on High-Performance Parallel and Distributed Computing, HPDC 2022 ; Conference date: 27-06-2022 Through 30-06-2022",

year = "2022",

month = jun,

day = "27",

doi = "10.1145/3502181.3531461",

language = "English",

series = "HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing",

publisher = "Association for Computing Machinery, Inc",

pages = "43--55",

booktitle = "HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing",

}

Bez, JL, Karimi, AM, Paul, AK, Xie, B, Byna, S, Carns, P, Oral, S , Wang, F & Hanley, J 2022, Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load. in HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing. HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing, Association for Computing Machinery, Inc, pp. 43-55, 31st International Symposium on High-Performance Parallel and Distributed Computing, HPDC 2022, Virtual, Online, United States, 06/27/22. https://doi.org/10.1145/3502181.3531461

Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load. / Bez, Jean Luca; Karimi, Ahmad Maroof; Paul, Arnab K. et al.
HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing. Association for Computing Machinery, Inc, 2022. p. 43-55 (HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing).

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

TY - GEN

T1 - Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load

AU - Bez, Jean Luca

AU - Karimi, Ahmad Maroof

AU - Paul, Arnab K.

AU - Xie, Bing

AU - Byna, Suren

AU - Carns, Philip

AU - Oral, Sarp

AU - Wang, Feiyi

AU - Hanley, Jesse

PY - 2022/6/27

Y1 - 2022/6/27

N2 - Scientific computing workloads at HPC facilities have been shifting from traditional numerical simulations to AI/ML applications for training and inference while processing and producing ever-increasing amounts of scientific data. To address the growing need for increased storage capacity, lower access latency, and higher bandwidth, emerging technologies such as non-volatile memory are integrated into supercomputer I/O subsystems. With these emerging trends, we need a better understanding of the multilayer supercomputer I/O systems and ways to use these subsystems efficiently. In this work, we study the I/O access patterns and performance characteristics of two representative supercomputer I/O subsystems. Through an extensive analysis of year-long I/O logs on each system, we report new observations in I/O reads and writes, unbalanced use of storage system layers, and new trends in user behaviors at the HPC I/O middleware stack.

AB - Scientific computing workloads at HPC facilities have been shifting from traditional numerical simulations to AI/ML applications for training and inference while processing and producing ever-increasing amounts of scientific data. To address the growing need for increased storage capacity, lower access latency, and higher bandwidth, emerging technologies such as non-volatile memory are integrated into supercomputer I/O subsystems. With these emerging trends, we need a better understanding of the multilayer supercomputer I/O systems and ways to use these subsystems efficiently. In this work, we study the I/O access patterns and performance characteristics of two representative supercomputer I/O subsystems. Through an extensive analysis of year-long I/O logs on each system, we report new observations in I/O reads and writes, unbalanced use of storage system layers, and new trends in user behaviors at the HPC I/O middleware stack.

KW - access patterns

KW - high-performance computing

KW - in-system storage

KW - parallel file systems

KW - production system

UR - http://www.scopus.com/inward/record.url?scp=85134187112&partnerID=8YFLogxK

U2 - 10.1145/3502181.3531461

DO - 10.1145/3502181.3531461

M3 - Conference contribution

AN - SCOPUS:85134187112

T3 - HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing

SP - 43

EP - 55

BT - HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing

PB - Association for Computing Machinery, Inc

T2 - 31st International Symposium on High-Performance Parallel and Distributed Computing, HPDC 2022

Y2 - 27 June 2022 through 30 June 2022

ER -

Bez JL, Karimi AM, Paul AK, Xie B, Byna S, Carns P et al. Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load. In HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing. Association for Computing Machinery, Inc. 2022. p. 43-55. (HPDC 2022 - Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing). doi: 10.1145/3502181.3531461

Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this