Linux block IO: introducing multi-queue SSD access on multi-core systems

Matias Bjørling; Jens Axboe; David Nellans; Philippe Bonnet

doi:10.1145/2485732.2485740

Linux block IO: introducing multi-queue SSD access on multi-core systems

Matias Bjørling, Jens Axboe, David Nellans, Philippe Bonnet

Computer Science

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

The IO performance of storage devices has accelerated from hundreds of IOPS five years ago, to hundreds of thousands of IOPS today, and tens of millions of IOPS projected in five years. This sharp evolution is primarily due to the introduc- tion of NAND-flash devices and their data parallel design. In this work, we demonstrate that the block layer within the operating system, originally designed to handle thousands of IOPS, has become a bottleneck to overall storage system performance, specially on the high NUMA-factor processors systems that are becoming commonplace. We describe the design of a next generation block layer that is capable of handling tens of millions of IOPS on a multi-core system equipped with a single storage device. Our experiments show that our design scales graciously with the number of cores, even on NUMA systems with multiple sockets.

Original language	English
Title of host publication	SYSTOR '13 Proceedings of the 6th International Systems and Storage Conference. Association for Computing Machinery, 22. (Systor).
Number of pages	1
Publisher	Association for Computing Machinery
Publication date	2013
Pages	1
Article number	22
ISBN (Print)	978-1-4503-2116-7
DOIs	https://doi.org/10.1145/2485732.2485740
Publication status	Published - 2013

Keywords

block layer
latency
linux
non-volatile memory
solid state drives
throughput

Access to Document

10.1145/2485732.2485740

http://dl.acm.org/citation.cfm?id=2485732.2485740

Cite this

@inproceedings{4ee19b4ee62445af921d017a368103f8,

title = "Linux block IO: introducing multi-queue SSD access on multi-core systems",

abstract = "The IO performance of storage devices has accelerated from hundreds of IOPS five years ago, to hundreds of thousands of IOPS today, and tens of millions of IOPS projected in five years. This sharp evolution is primarily due to the introduc- tion of NAND-flash devices and their data parallel design. In this work, we demonstrate that the block layer within the operating system, originally designed to handle thousands of IOPS, has become a bottleneck to overall storage system performance, specially on the high NUMA-factor processors systems that are becoming commonplace. We describe the design of a next generation block layer that is capable of handling tens of millions of IOPS on a multi-core system equipped with a single storage device. Our experiments show that our design scales graciously with the number of cores, even on NUMA systems with multiple sockets.",

keywords = "block layer, latency, linux, non-volatile memory, solid state drives, throughput",

author = "Matias Bj{\o}rling and Jens Axboe and David Nellans and Philippe Bonnet",

year = "2013",

doi = "10.1145/2485732.2485740",

language = "English",

isbn = "978-1-4503-2116-7 ",

pages = "1",

booktitle = "SYSTOR '13 Proceedings of the 6th International Systems and Storage Conference. Association for Computing Machinery, 22. (Systor).",

publisher = "Association for Computing Machinery",

address = "United States",

}

Linux block IO: introducing multi-queue SSD access on multi-core systems. / Bjørling, Matias; Axboe, Jens; Nellans, David et al.
SYSTOR '13 Proceedings of the 6th International Systems and Storage Conference. Association for Computing Machinery, 22. (Systor).. Association for Computing Machinery, 2013. p. 1 22.

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

TY - GEN

T1 - Linux block IO: introducing multi-queue SSD access on multi-core systems

AU - Bjørling, Matias

AU - Axboe, Jens

AU - Nellans, David

AU - Bonnet, Philippe

PY - 2013

Y1 - 2013

N2 - The IO performance of storage devices has accelerated from hundreds of IOPS five years ago, to hundreds of thousands of IOPS today, and tens of millions of IOPS projected in five years. This sharp evolution is primarily due to the introduc- tion of NAND-flash devices and their data parallel design. In this work, we demonstrate that the block layer within the operating system, originally designed to handle thousands of IOPS, has become a bottleneck to overall storage system performance, specially on the high NUMA-factor processors systems that are becoming commonplace. We describe the design of a next generation block layer that is capable of handling tens of millions of IOPS on a multi-core system equipped with a single storage device. Our experiments show that our design scales graciously with the number of cores, even on NUMA systems with multiple sockets.

AB - The IO performance of storage devices has accelerated from hundreds of IOPS five years ago, to hundreds of thousands of IOPS today, and tens of millions of IOPS projected in five years. This sharp evolution is primarily due to the introduc- tion of NAND-flash devices and their data parallel design. In this work, we demonstrate that the block layer within the operating system, originally designed to handle thousands of IOPS, has become a bottleneck to overall storage system performance, specially on the high NUMA-factor processors systems that are becoming commonplace. We describe the design of a next generation block layer that is capable of handling tens of millions of IOPS on a multi-core system equipped with a single storage device. Our experiments show that our design scales graciously with the number of cores, even on NUMA systems with multiple sockets.

KW - block layer

KW - latency

KW - linux

KW - non-volatile memory

KW - solid state drives

KW - throughput

U2 - 10.1145/2485732.2485740

DO - 10.1145/2485732.2485740

M3 - Article in proceedings

SN - 978-1-4503-2116-7

SP - 1

BT - SYSTOR '13 Proceedings of the 6th International Systems and Storage Conference. Association for Computing Machinery, 22. (Systor).

PB - Association for Computing Machinery

ER -

Linux block IO: introducing multi-queue SSD access on multi-core systems

Abstract

Keywords

Access to Document

Fingerprint

Cite this