Net-cohort: Detecting and managing VM ensembles in virtualized data centers

Liting Hu; Karsten Schwan; Ajay Gulati; Junjie Zhang; Chengwei Wang

doi:10.1145/2371536.2371540

Net-cohort: Detecting and managing VM ensembles in virtualized data centers

Liting Hu, Karsten Schwan, Ajay Gulati, Junjie Zhang, Chengwei Wang

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

Abstract

Bi-section bandwidth is a critical resource in today’s data centers because of the high cost and limited bandwidth of higher-level network switches and routers. This problem is aggravated in virtualized environments where a set of virtual machines, jointly implementing some service, may run across multiple L2 hops. Since data center administrators typically do not have visibility into such sets of communicating VMs, this can cause inter-VM traffic to traverse bottlenecked network paths. To address this problem, we present `Net-Cohort’, which offers lightweight system-level techniques to (1) discover VM ensembles and (2) collect information about intra-ensemble VM interactions. Net- Cohort can dynamically identify ensembles to manipulate entire services/applications rather than individual VMs, and to support VM placement engines in co-locating communicating VMs in order to reduce the consumption of bi-section bandwidth. An implementation of Net-Cohort on a Xen-based system with 15 hosts and 225 VMs shows that its methods can detect VM ensembles at low cost and with about 90.0% accuracy. Placements based on ensemble information provided by Net- Cohort can result in an up to 385% improvement in application throughput for a RUBiS instance, a 56.4% improvement in application throughput for a Hadoop instance, and a 12.76 times improvement in quality of service for a SIPp instance.

Original language	American English
Title of host publication	ICAC '12: Proceedings of the 9th international conference on Autonomic computing
Publisher	Publ by ACM
Pages	3-12
Number of pages	10
ISBN (Print)	9781450315203
DOIs	https://doi.org/10.1145/2371536.2371540
State	Published - Sep 1 2012
Externally published	Yes
Event	9th ACM International Conference on Autonomic Computing, ICAC'12 - San Jose, CA, United States Duration: Sep 18 2012 → Sep 20 2012

Conference

Conference	9th ACM International Conference on Autonomic Computing, ICAC'12
Country/Territory	United States
City	San Jose, CA
Period	9/18/12 → 9/20/12

ASJC Scopus Subject Areas

Artificial Intelligence
Applied Mathematics

Keywords

Algorithms
Clustering
Dependency Analysis
Design
Management
Virtualization

Disciplines

Computer Sciences
Engineering

Access to Document

10.1145/2371536.2371540License: Unspecified

Cite this

Hu, L, Schwan, K, Gulati, A, Zhang, J & Wang, C 2012, Net-cohort: Detecting and managing VM ensembles in virtualized data centers. in ICAC '12: Proceedings of the 9th international conference on Autonomic computing. Publ by ACM, pp. 3-12, 9th ACM International Conference on Autonomic Computing, ICAC'12, San Jose, CA, United States, 9/18/12. https://doi.org/10.1145/2371536.2371540

@inproceedings{58cff5d2e222438fa54efe7b739d35ec,

title = "Net-cohort: Detecting and managing VM ensembles in virtualized data centers",

abstract = " Bi-section bandwidth is a critical resource in today{\textquoteright}s data centers because of the high cost and limited bandwidth of higher-level network switches and routers. This problem is aggravated in virtualized environments where a set of virtual machines, jointly implementing some service, may run across multiple L2 hops. Since data center administrators typically do not have visibility into such sets of communicating VMs, this can cause inter-VM traffic to traverse bottlenecked network paths. To address this problem, we present `Net-Cohort{\textquoteright}, which offers lightweight system-level techniques to (1) discover VM ensembles and (2) collect information about intra-ensemble VM interactions. Net- Cohort can dynamically identify ensembles to manipulate entire services/applications rather than individual VMs, and to support VM placement engines in co-locating communicating VMs in order to reduce the consumption of bi-section bandwidth. An implementation of Net-Cohort on a Xen-based system with 15 hosts and 225 VMs shows that its methods can detect VM ensembles at low cost and with about 90.0% accuracy. Placements based on ensemble information provided by Net- Cohort can result in an up to 385% improvement in application throughput for a RUBiS instance, a 56.4% improvement in application throughput for a Hadoop instance, and a 12.76 times improvement in quality of service for a SIPp instance.",

keywords = "Algorithms, Clustering, Dependency Analysis, Design, Management, Virtualization",

author = "Liting Hu and Karsten Schwan and Ajay Gulati and Junjie Zhang and Chengwei Wang",

year = "2012",

month = sep,

day = "1",

doi = "10.1145/2371536.2371540",

language = "American English",

isbn = "9781450315203",

pages = "3--12",

booktitle = "ICAC '12: Proceedings of the 9th international conference on Autonomic computing",

publisher = "Publ by ACM",

note = "9th ACM International Conference on Autonomic Computing, ICAC'12 ; Conference date: 18-09-2012 Through 20-09-2012",

}

TY - GEN

T1 - Net-cohort

T2 - 9th ACM International Conference on Autonomic Computing, ICAC'12

AU - Hu, Liting

AU - Schwan, Karsten

AU - Gulati, Ajay

AU - Zhang, Junjie

AU - Wang, Chengwei

PY - 2012/9/1

Y1 - 2012/9/1

N2 - Bi-section bandwidth is a critical resource in today’s data centers because of the high cost and limited bandwidth of higher-level network switches and routers. This problem is aggravated in virtualized environments where a set of virtual machines, jointly implementing some service, may run across multiple L2 hops. Since data center administrators typically do not have visibility into such sets of communicating VMs, this can cause inter-VM traffic to traverse bottlenecked network paths. To address this problem, we present `Net-Cohort’, which offers lightweight system-level techniques to (1) discover VM ensembles and (2) collect information about intra-ensemble VM interactions. Net- Cohort can dynamically identify ensembles to manipulate entire services/applications rather than individual VMs, and to support VM placement engines in co-locating communicating VMs in order to reduce the consumption of bi-section bandwidth. An implementation of Net-Cohort on a Xen-based system with 15 hosts and 225 VMs shows that its methods can detect VM ensembles at low cost and with about 90.0% accuracy. Placements based on ensemble information provided by Net- Cohort can result in an up to 385% improvement in application throughput for a RUBiS instance, a 56.4% improvement in application throughput for a Hadoop instance, and a 12.76 times improvement in quality of service for a SIPp instance.

AB - Bi-section bandwidth is a critical resource in today’s data centers because of the high cost and limited bandwidth of higher-level network switches and routers. This problem is aggravated in virtualized environments where a set of virtual machines, jointly implementing some service, may run across multiple L2 hops. Since data center administrators typically do not have visibility into such sets of communicating VMs, this can cause inter-VM traffic to traverse bottlenecked network paths. To address this problem, we present `Net-Cohort’, which offers lightweight system-level techniques to (1) discover VM ensembles and (2) collect information about intra-ensemble VM interactions. Net- Cohort can dynamically identify ensembles to manipulate entire services/applications rather than individual VMs, and to support VM placement engines in co-locating communicating VMs in order to reduce the consumption of bi-section bandwidth. An implementation of Net-Cohort on a Xen-based system with 15 hosts and 225 VMs shows that its methods can detect VM ensembles at low cost and with about 90.0% accuracy. Placements based on ensemble information provided by Net- Cohort can result in an up to 385% improvement in application throughput for a RUBiS instance, a 56.4% improvement in application throughput for a Hadoop instance, and a 12.76 times improvement in quality of service for a SIPp instance.

KW - Algorithms

KW - Clustering

KW - Dependency Analysis

KW - Design

KW - Management

KW - Virtualization

UR - http://www.scopus.com/inward/record.url?scp=84867695892&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867695892&partnerID=8YFLogxK

UR - https://corescholar.libraries.wright.edu/cse/8

U2 - 10.1145/2371536.2371540

DO - 10.1145/2371536.2371540

M3 - Conference contribution

SN - 9781450315203

SP - 3

EP - 12

BT - ICAC '12: Proceedings of the 9th international conference on Autonomic computing

PB - Publ by ACM

Y2 - 18 September 2012 through 20 September 2012

ER -

Net-cohort: Detecting and managing VM ensembles in virtualized data centers

Abstract

Conference

ASJC Scopus Subject Areas

Keywords

Disciplines

Access to Document

Other files and links

Cite this