UFLIP-OC: Understanding flash I/O patterns on open-channel solid-state drives

Ivan Luiz Picoli; Carla Villegas Pasco; Björn Pór Jónsson; Luc Bouganim; Philippe Bonnet

doi:10.1145/3124680.3124741

UFLIP-OC: Understanding flash I/O patterns on open-channel solid-state drives

Ivan Luiz Picoli, Carla Villegas Pasco, Björn Pór Jónsson, Luc Bouganim, Philippe Bonnet

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

11 Citations (Scopus)

Abstract

Solid-State Drives (SSDs) have gained acceptance by providing the same block device abstraction as magnetic hard drives, at the cost of suboptimal resource utilisation and unpredictable performance. Recently, Open-Channel SSDs have emerged as a means to obtain predictably high performance, based on a clean break from the block device abstraction. Open-channel SSDs embed a minimal flash translation layer (FTL) and expose their internals to the host. The Linux open-channel SSD subsystem, LightNVM, lets kernel modules as well as user-space applications control data placement and I/O scheduling. This way, it is the host that is responsible for SSD management. But what kind of performance model should the host rely on to guide the way it manages data placement and I/O scheduling? For addressing this question we have defined uFLIPOC, a benchmark designed to identify the I/O patterns that are best suited for a given open-channel SSD. Our experiments on a Dragon- Fire Card (DFC) SSD, equipped with the OX controller, illustrate the performance impact of media characteristics and parallelism. We discuss how uFLIP-OC can be used to guide the design of host-based data systems on open-channel SSDs.

Original language	English
Title of host publication	Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017
Publisher	Association for Computing Machinery, Inc.
Publication date	2 Sep 2017
Article number	3124741
ISBN (Electronic)	9781450351973
DOIs	https://doi.org/10.1145/3124680.3124741
Publication status	Published - 2 Sep 2017
Externally published	Yes
Event	8th ACM Asia Pacific Conference on Systems, APSys 2017 - Mumbai, India Duration: 2 Sep 2017 → …

Conference

Conference	8th ACM Asia Pacific Conference on Systems, APSys 2017
Country/Territory	India
City	Mumbai
Period	02/09/2017 → …
Sponsor	ACM Special Interest Group on Operating Systems (SIGOPS)

Series	Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017

Bibliographical note

Funding Information:
This research was partially supported by the Coordination for the Improvement of Higher Education Personnel (CAPES), Brazil, who provided Ph.D fellowship for the first author.

Keywords

Benchmarking
NAND flash
Open-channel SSDs
uFLIP-OC

Access to Document

10.1145/3124680.3124741

Cite this

Picoli, I. L., Pasco, C. V., Jónsson, B. P., Bouganim, L., & Bonnet, P. (2017). UFLIP-OC: Understanding flash I/O patterns on open-channel solid-state drives. In Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017 [3124741] Association for Computing Machinery, Inc.. Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017 https://doi.org/10.1145/3124680.3124741

UFLIP-OC : Understanding flash I/O patterns on open-channel solid-state drives. / Picoli, Ivan Luiz; Pasco, Carla Villegas; Jónsson, Björn Pór; Bouganim, Luc; Bonnet, Philippe.

Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017. Association for Computing Machinery, Inc., 2017. 3124741 (Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017).

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

Picoli, IL, Pasco, CV, Jónsson, BP, Bouganim, L & Bonnet, P 2017, UFLIP-OC: Understanding flash I/O patterns on open-channel solid-state drives. in Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017., 3124741, Association for Computing Machinery, Inc., Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017, 8th ACM Asia Pacific Conference on Systems, APSys 2017, Mumbai, India, 02/09/2017. https://doi.org/10.1145/3124680.3124741

Picoli IL, Pasco CV, Jónsson BP, Bouganim L, Bonnet P. UFLIP-OC: Understanding flash I/O patterns on open-channel solid-state drives. In Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017. Association for Computing Machinery, Inc. 2017. 3124741. (Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017). https://doi.org/10.1145/3124680.3124741

Picoli, Ivan Luiz ; Pasco, Carla Villegas ; Jónsson, Björn Pór ; Bouganim, Luc ; Bonnet, Philippe. / UFLIP-OC : Understanding flash I/O patterns on open-channel solid-state drives. Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017. Association for Computing Machinery, Inc., 2017. (Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017).

@inproceedings{1112f90f08f8438fb5837ae34d74b50f,

title = "UFLIP-OC: Understanding flash I/O patterns on open-channel solid-state drives",

abstract = "Solid-State Drives (SSDs) have gained acceptance by providing the same block device abstraction as magnetic hard drives, at the cost of suboptimal resource utilisation and unpredictable performance. Recently, Open-Channel SSDs have emerged as a means to obtain predictably high performance, based on a clean break from the block device abstraction. Open-channel SSDs embed a minimal flash translation layer (FTL) and expose their internals to the host. The Linux open-channel SSD subsystem, LightNVM, lets kernel modules as well as user-space applications control data placement and I/O scheduling. This way, it is the host that is responsible for SSD management. But what kind of performance model should the host rely on to guide the way it manages data placement and I/O scheduling? For addressing this question we have defined uFLIPOC, a benchmark designed to identify the I/O patterns that are best suited for a given open-channel SSD. Our experiments on a Dragon- Fire Card (DFC) SSD, equipped with the OX controller, illustrate the performance impact of media characteristics and parallelism. We discuss how uFLIP-OC can be used to guide the design of host-based data systems on open-channel SSDs.",

keywords = "Benchmarking, NAND flash, Open-channel SSDs, uFLIP-OC",

author = "Picoli, {Ivan Luiz} and Pasco, {Carla Villegas} and J{\'o}nsson, {Bj{\"o}rn P{\'o}r} and Luc Bouganim and Philippe Bonnet",

note = "Funding Information: This research was partially supported by the Coordination for the Improvement of Higher Education Personnel (CAPES), Brazil, who provided Ph.D fellowship for the first author.; 8th ACM Asia Pacific Conference on Systems, APSys 2017 ; Conference date: 02-09-2017",

year = "2017",

month = sep,

day = "2",

doi = "10.1145/3124680.3124741",

language = "English",

series = "Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017",

booktitle = "Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017",

publisher = "Association for Computing Machinery, Inc.",

}

TY - GEN

T1 - UFLIP-OC

T2 - 8th ACM Asia Pacific Conference on Systems, APSys 2017

AU - Picoli, Ivan Luiz

AU - Pasco, Carla Villegas

AU - Jónsson, Björn Pór

AU - Bouganim, Luc

AU - Bonnet, Philippe

N1 - Funding Information: This research was partially supported by the Coordination for the Improvement of Higher Education Personnel (CAPES), Brazil, who provided Ph.D fellowship for the first author.

PY - 2017/9/2

Y1 - 2017/9/2

N2 - Solid-State Drives (SSDs) have gained acceptance by providing the same block device abstraction as magnetic hard drives, at the cost of suboptimal resource utilisation and unpredictable performance. Recently, Open-Channel SSDs have emerged as a means to obtain predictably high performance, based on a clean break from the block device abstraction. Open-channel SSDs embed a minimal flash translation layer (FTL) and expose their internals to the host. The Linux open-channel SSD subsystem, LightNVM, lets kernel modules as well as user-space applications control data placement and I/O scheduling. This way, it is the host that is responsible for SSD management. But what kind of performance model should the host rely on to guide the way it manages data placement and I/O scheduling? For addressing this question we have defined uFLIPOC, a benchmark designed to identify the I/O patterns that are best suited for a given open-channel SSD. Our experiments on a Dragon- Fire Card (DFC) SSD, equipped with the OX controller, illustrate the performance impact of media characteristics and parallelism. We discuss how uFLIP-OC can be used to guide the design of host-based data systems on open-channel SSDs.

AB - Solid-State Drives (SSDs) have gained acceptance by providing the same block device abstraction as magnetic hard drives, at the cost of suboptimal resource utilisation and unpredictable performance. Recently, Open-Channel SSDs have emerged as a means to obtain predictably high performance, based on a clean break from the block device abstraction. Open-channel SSDs embed a minimal flash translation layer (FTL) and expose their internals to the host. The Linux open-channel SSD subsystem, LightNVM, lets kernel modules as well as user-space applications control data placement and I/O scheduling. This way, it is the host that is responsible for SSD management. But what kind of performance model should the host rely on to guide the way it manages data placement and I/O scheduling? For addressing this question we have defined uFLIPOC, a benchmark designed to identify the I/O patterns that are best suited for a given open-channel SSD. Our experiments on a Dragon- Fire Card (DFC) SSD, equipped with the OX controller, illustrate the performance impact of media characteristics and parallelism. We discuss how uFLIP-OC can be used to guide the design of host-based data systems on open-channel SSDs.

KW - Benchmarking

KW - NAND flash

KW - Open-channel SSDs

KW - uFLIP-OC

U2 - 10.1145/3124680.3124741

DO - 10.1145/3124680.3124741

M3 - Article in proceedings

AN - SCOPUS:85030566055

T3 - Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017

BT - Proceedings of the 8th Asia-Pacific Workshop on Systems, APSys 2017

PB - Association for Computing Machinery, Inc.

Y2 - 2 September 2017

ER -