Provisioning ZFS pools on lustre

Rick Mohr; Adam P. Howard

doi:10.1145/3332186.3332245

Provisioning ZFS pools on lustre

Rick Mohr, Adam P. Howard

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

Abstract

While Lustre’s parallelism, performance, and scalability make it desirable as a storage solution for clusters, its limitations prevent it from being suitable as a general purpose storage for all of a cluster’s needs. In particular, Lustre’s relatively poor performance for small, random I/O as well as metadata-intensive workloads make it less suitable for use as storage for users’ home areas or working directories for compilation. In this paper, an experiment of deploying a ZFS file system using files pre-allocated on a Lustre file system as the ZFS storage pools is presented. While this adds many management options like snapshots and more fine grained controls over users and quotas, we focus on examining how adding this layer can affect the performance of suboptimal Lustre workloads. Benchmarking of the Lustre file system with and without the ZFS file system is performed, and the results are presented and analyzed.

Original language	English
Title of host publication	Proceedings of the Practice and Experience in Advanced Research Computing
Subtitle of host publication	Rise of the Machines (Learning), PEARC 2019
Publisher	Association for Computing Machinery
ISBN (Electronic)	9781450372275
DOIs	https://doi.org/10.1145/3332186.3332245
State	Published - Jul 28 2019
Externally published	Yes
Event	2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019 - Chicago, United States Duration: Jul 28 2019 → Aug 1 2019

Publication series

Name	ACM International Conference Proceeding Series

Conference

Conference	2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019
Country/Territory	United States
City	Chicago
Period	07/28/19 → 08/1/19

Keywords

Benchmarking
Lustre
ZFS

Access to Document

10.1145/3332186.3332245

Cite this

@inproceedings{6dc645398e994866bb8cfc8dd9e8b749,

title = "Provisioning ZFS pools on lustre",

abstract = "While Lustre{\textquoteright}s parallelism, performance, and scalability make it desirable as a storage solution for clusters, its limitations prevent it from being suitable as a general purpose storage for all of a cluster{\textquoteright}s needs. In particular, Lustre{\textquoteright}s relatively poor performance for small, random I/O as well as metadata-intensive workloads make it less suitable for use as storage for users{\textquoteright} home areas or working directories for compilation. In this paper, an experiment of deploying a ZFS file system using files pre-allocated on a Lustre file system as the ZFS storage pools is presented. While this adds many management options like snapshots and more fine grained controls over users and quotas, we focus on examining how adding this layer can affect the performance of suboptimal Lustre workloads. Benchmarking of the Lustre file system with and without the ZFS file system is performed, and the results are presented and analyzed.",

keywords = "Benchmarking, Lustre, ZFS",

author = "Rick Mohr and Howard, {Adam P.}",

note = "Publisher Copyright: {\textcopyright} 2019 Association for Computing Machinery.; 2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019 ; Conference date: 28-07-2019 Through 01-08-2019",

year = "2019",

month = jul,

day = "28",

doi = "10.1145/3332186.3332245",

language = "English",

series = "ACM International Conference Proceeding Series",

publisher = "Association for Computing Machinery",

booktitle = "Proceedings of the Practice and Experience in Advanced Research Computing",

}

Mohr, R & Howard, AP 2019, Provisioning ZFS pools on lustre. in Proceedings of the Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019., 3332245, ACM International Conference Proceeding Series, Association for Computing Machinery, 2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019, Chicago, United States, 07/28/19. https://doi.org/10.1145/3332186.3332245

Provisioning ZFS pools on lustre. / Mohr, Rick; Howard, Adam P.
Proceedings of the Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019. Association for Computing Machinery, 2019. 3332245 (ACM International Conference Proceeding Series).

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

TY - GEN

T1 - Provisioning ZFS pools on lustre

AU - Mohr, Rick

AU - Howard, Adam P.

PY - 2019/7/28

Y1 - 2019/7/28

N2 - While Lustre’s parallelism, performance, and scalability make it desirable as a storage solution for clusters, its limitations prevent it from being suitable as a general purpose storage for all of a cluster’s needs. In particular, Lustre’s relatively poor performance for small, random I/O as well as metadata-intensive workloads make it less suitable for use as storage for users’ home areas or working directories for compilation. In this paper, an experiment of deploying a ZFS file system using files pre-allocated on a Lustre file system as the ZFS storage pools is presented. While this adds many management options like snapshots and more fine grained controls over users and quotas, we focus on examining how adding this layer can affect the performance of suboptimal Lustre workloads. Benchmarking of the Lustre file system with and without the ZFS file system is performed, and the results are presented and analyzed.

AB - While Lustre’s parallelism, performance, and scalability make it desirable as a storage solution for clusters, its limitations prevent it from being suitable as a general purpose storage for all of a cluster’s needs. In particular, Lustre’s relatively poor performance for small, random I/O as well as metadata-intensive workloads make it less suitable for use as storage for users’ home areas or working directories for compilation. In this paper, an experiment of deploying a ZFS file system using files pre-allocated on a Lustre file system as the ZFS storage pools is presented. While this adds many management options like snapshots and more fine grained controls over users and quotas, we focus on examining how adding this layer can affect the performance of suboptimal Lustre workloads. Benchmarking of the Lustre file system with and without the ZFS file system is performed, and the results are presented and analyzed.

KW - Benchmarking

KW - Lustre

KW - ZFS

UR - http://www.scopus.com/inward/record.url?scp=85071032636&partnerID=8YFLogxK

U2 - 10.1145/3332186.3332245

DO - 10.1145/3332186.3332245

M3 - Conference contribution

AN - SCOPUS:85071032636

T3 - ACM International Conference Proceeding Series

BT - Proceedings of the Practice and Experience in Advanced Research Computing

PB - Association for Computing Machinery

T2 - 2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019

Y2 - 28 July 2019 through 1 August 2019

ER -

Provisioning ZFS pools on lustre

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this