Benchmarking Parallelism in Unikernels

Akilan Selvacoumar; Robert James Stewart; Hans-Wolfgang Loidl; Ahmad Ryad Soobhany

doi:10.1007/978-3-031-31209-0_23

Benchmarking Parallelism in Unikernels

Akilan Selvacoumar^*, Robert James Stewart, Hans-Wolfgang Loidl, Ahmad Ryad Soobhany

^*Corresponding author for this work

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

21 Downloads (Pure)

Abstract

Virtualisation technologies are widely used in Cloud computing infrastructures, because they can be provisioned cheaply and quickly to meet demand. The common approaches are either to package a Operating System (OS) as a Virtual Machine, or to containerise software with an OS kernel. An emerging alternative are unikernels, which are customised kernels to support just one application. Unikernels are lightweight and an applications has sole use of the kernel, which offers potential for fast, resource efficient and secure execution. For these reasons, unikernels may be idea for parallel computing in the Cloud. However, the parallel performance of unikernel-based Cloud applications has not been extensively studied. This paper presents an evaluation of the OSv unikernel using a parallelised Mandelbrot benchmark, comparing with Docker and a monolithic VM for runtime, parallel speedups and boot-up time. OSv has the fastest boot-up time, and is comparable with the parallel speedups of Docker and the monolithic VM.

Original language	English
Title of host publication	Euro-Par 2022: Parallel Processing Workshops
Publisher	Springer
Pages	286–293
Number of pages	8
ISBN (Electronic)	9783031312090
ISBN (Print)	9783031312083
DOIs	https://doi.org/10.1007/978-3-031-31209-0_23
Publication status	Published - 2 May 2023
Event	European Conference on Parallel Processing: PhD Symposium - Glasgow, United Kingdom Duration: 22 Aug 2022 → 26 Aug 2022 https://link.springer.com/book/10.1007/978-3-031-31209-0

Publication series

Name	Lecture Notes in Computer Science
Volume	13835
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	European Conference on Parallel Processing
Abbreviated title	EuroPar PhD Symposium
Country/Territory	United Kingdom
City	Glasgow
Period	22/08/22 → 26/08/22
Internet address	https://link.springer.com/book/10.1007/978-3-031-31209-0

Keywords

Parallel Computing
Unikernels

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-031-31209-0_23

Download pdf
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG.
Accepted author manuscript, 3.99 MB

Cite this

@inproceedings{0b3fcd9e40714a4f9a831b481f727b03,

title = "Benchmarking Parallelism in Unikernels",

abstract = "Virtualisation technologies are widely used in Cloud computing infrastructures, because they can be provisioned cheaply and quickly to meet demand. The common approaches are either to package a Operating System (OS) as a Virtual Machine, or to containerise software with an OS kernel. An emerging alternative are unikernels, which are customised kernels to support just one application. Unikernels are lightweight and an applications has sole use of the kernel, which offers potential for fast, resource efficient and secure execution. For these reasons, unikernels may be idea for parallel computing in the Cloud. However, the parallel performance of unikernel-based Cloud applications has not been extensively studied. This paper presents an evaluation of the OSv unikernel using a parallelised Mandelbrot benchmark, comparing with Docker and a monolithic VM for runtime, parallel speedups and boot-up time. OSv has the fastest boot-up time, and is comparable with the parallel speedups of Docker and the monolithic VM.",

keywords = "Parallel Computing, Unikernels",

author = "Akilan Selvacoumar and Stewart, {Robert James} and Hans-Wolfgang Loidl and Soobhany, {Ahmad Ryad}",

year = "2023",

month = may,

day = "2",

doi = "10.1007/978-3-031-31209-0_23",

language = "English",

isbn = "9783031312083",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "286–293",

booktitle = "Euro-Par 2022: Parallel Processing Workshops",

note = "European Conference on Parallel Processing : PhD Symposium, EuroPar PhD Symposium ; Conference date: 22-08-2022 Through 26-08-2022",

url = "https://link.springer.com/book/10.1007/978-3-031-31209-0",

}

TY - GEN

T1 - Benchmarking Parallelism in Unikernels

AU - Selvacoumar, Akilan

AU - Stewart, Robert James

AU - Loidl, Hans-Wolfgang

AU - Soobhany, Ahmad Ryad

PY - 2023/5/2

Y1 - 2023/5/2

N2 - Virtualisation technologies are widely used in Cloud computing infrastructures, because they can be provisioned cheaply and quickly to meet demand. The common approaches are either to package a Operating System (OS) as a Virtual Machine, or to containerise software with an OS kernel. An emerging alternative are unikernels, which are customised kernels to support just one application. Unikernels are lightweight and an applications has sole use of the kernel, which offers potential for fast, resource efficient and secure execution. For these reasons, unikernels may be idea for parallel computing in the Cloud. However, the parallel performance of unikernel-based Cloud applications has not been extensively studied. This paper presents an evaluation of the OSv unikernel using a parallelised Mandelbrot benchmark, comparing with Docker and a monolithic VM for runtime, parallel speedups and boot-up time. OSv has the fastest boot-up time, and is comparable with the parallel speedups of Docker and the monolithic VM.

AB - Virtualisation technologies are widely used in Cloud computing infrastructures, because they can be provisioned cheaply and quickly to meet demand. The common approaches are either to package a Operating System (OS) as a Virtual Machine, or to containerise software with an OS kernel. An emerging alternative are unikernels, which are customised kernels to support just one application. Unikernels are lightweight and an applications has sole use of the kernel, which offers potential for fast, resource efficient and secure execution. For these reasons, unikernels may be idea for parallel computing in the Cloud. However, the parallel performance of unikernel-based Cloud applications has not been extensively studied. This paper presents an evaluation of the OSv unikernel using a parallelised Mandelbrot benchmark, comparing with Docker and a monolithic VM for runtime, parallel speedups and boot-up time. OSv has the fastest boot-up time, and is comparable with the parallel speedups of Docker and the monolithic VM.

KW - Parallel Computing

KW - Unikernels

UR - http://www.scopus.com/inward/record.url?scp=85161399979&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-31209-0_23

DO - 10.1007/978-3-031-31209-0_23

M3 - Conference contribution

SN - 9783031312083

T3 - Lecture Notes in Computer Science

SP - 286

EP - 293

BT - Euro-Par 2022: Parallel Processing Workshops

PB - Springer

T2 - European Conference on Parallel Processing

Y2 - 22 August 2022 through 26 August 2022

ER -

Benchmarking Parallelism in Unikernels

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this