Inferring traffic burstiness by sampling the buffer occupancy

Michel Mandjes, Remco van de Meent

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

2 Citations (Scopus)

Abstract

Common practice to determine the required bandwidth capacity for a network link is to measure the 5 minute average link load, and then add a safety margin to cater for the effect of burstiness on small time-scales. Because of the substantial measurement efforts required to determine the burstiness, network managers often rely on rules of thumb to find the safety margin, e.g. 'mean plus 50%'. In this paper we propose a novel method to accurately determine the burstiness of traffic on small time-scales, without requiring measurements on such small time-scales. Our method is based on coarse-grained polling of the occupancy of a buffer in front of the link, from which the burstiness on small timescales is inferred. We provide the theoretical foundations of our approach, and a validation through both simulation using synthetic traffic as well as real network traffic taken from various operational networks. It turns out that using our approach, it is possible to accurately determine burstiness on small time-scales (for instance 10 ms), by sampling the buffer occupancy (for instance) every second.
Original languageEnglish
Title of host publicationNETWORKING 2005
Subtitle of host publication4th International IFIP-TC6 Networking Conference, Waterloo, Canada, May 2-6, 2005. Proceedings
EditorsR. Boutaba, K.C. Almeroth, R. Puigjaner, S. Shen, J.P. Black
Place of PublicationBerlin
PublisherSpringer
Pages303-315
Number of pages13
ISBN (Electronic)978-3-540-32017-3
ISBN (Print)978-3-540-25809-4
DOIs
Publication statusPublished - 2005
Event4th International IFIP-TC6 Networking Conference, NETWORKING 2005 - Waterloo, Canada
Duration: 2 May 20056 May 2005
Conference number: 4

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume3462
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference4th International IFIP-TC6 Networking Conference, NETWORKING 2005
Abbreviated titleNETWORKING
CountryCanada
CityWaterloo
Period2/05/056/05/05

Fingerprint

Sampling
Telecommunication links
Managers
Bandwidth

Keywords

  • Network design and capacity planning
  • queueing network models
  • IR-63468
  • METIS-225554
  • EWI-7069

Cite this

Mandjes, M., & van de Meent, R. (2005). Inferring traffic burstiness by sampling the buffer occupancy. In R. Boutaba, K. C. Almeroth, R. Puigjaner, S. Shen, & J. P. Black (Eds.), NETWORKING 2005: 4th International IFIP-TC6 Networking Conference, Waterloo, Canada, May 2-6, 2005. Proceedings (pp. 303-315). (Lecture Notes in Computer Science; Vol. 3462). Berlin: Springer. https://doi.org/10.1007/11422778_25
Mandjes, Michel ; van de Meent, Remco. / Inferring traffic burstiness by sampling the buffer occupancy. NETWORKING 2005: 4th International IFIP-TC6 Networking Conference, Waterloo, Canada, May 2-6, 2005. Proceedings. editor / R. Boutaba ; K.C. Almeroth ; R. Puigjaner ; S. Shen ; J.P. Black. Berlin : Springer, 2005. pp. 303-315 (Lecture Notes in Computer Science).
@inproceedings{d37d1ea6765648e289245fd0392ba649,
title = "Inferring traffic burstiness by sampling the buffer occupancy",
abstract = "Common practice to determine the required bandwidth capacity for a network link is to measure the 5 minute average link load, and then add a safety margin to cater for the effect of burstiness on small time-scales. Because of the substantial measurement efforts required to determine the burstiness, network managers often rely on rules of thumb to find the safety margin, e.g. 'mean plus 50{\%}'. In this paper we propose a novel method to accurately determine the burstiness of traffic on small time-scales, without requiring measurements on such small time-scales. Our method is based on coarse-grained polling of the occupancy of a buffer in front of the link, from which the burstiness on small timescales is inferred. We provide the theoretical foundations of our approach, and a validation through both simulation using synthetic traffic as well as real network traffic taken from various operational networks. It turns out that using our approach, it is possible to accurately determine burstiness on small time-scales (for instance 10 ms), by sampling the buffer occupancy (for instance) every second.",
keywords = "Network design and capacity planning, queueing network models, IR-63468, METIS-225554, EWI-7069",
author = "Michel Mandjes and {van de Meent}, Remco",
year = "2005",
doi = "10.1007/11422778_25",
language = "English",
isbn = "978-3-540-25809-4",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "303--315",
editor = "R. Boutaba and K.C. Almeroth and R. Puigjaner and S. Shen and J.P. Black",
booktitle = "NETWORKING 2005",

}

Mandjes, M & van de Meent, R 2005, Inferring traffic burstiness by sampling the buffer occupancy. in R Boutaba, KC Almeroth, R Puigjaner, S Shen & JP Black (eds), NETWORKING 2005: 4th International IFIP-TC6 Networking Conference, Waterloo, Canada, May 2-6, 2005. Proceedings. Lecture Notes in Computer Science, vol. 3462, Springer, Berlin, pp. 303-315, 4th International IFIP-TC6 Networking Conference, NETWORKING 2005, Waterloo, Canada, 2/05/05. https://doi.org/10.1007/11422778_25

Inferring traffic burstiness by sampling the buffer occupancy. / Mandjes, Michel; van de Meent, Remco.

NETWORKING 2005: 4th International IFIP-TC6 Networking Conference, Waterloo, Canada, May 2-6, 2005. Proceedings. ed. / R. Boutaba; K.C. Almeroth; R. Puigjaner; S. Shen; J.P. Black. Berlin : Springer, 2005. p. 303-315 (Lecture Notes in Computer Science; Vol. 3462).

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Inferring traffic burstiness by sampling the buffer occupancy

AU - Mandjes, Michel

AU - van de Meent, Remco

PY - 2005

Y1 - 2005

N2 - Common practice to determine the required bandwidth capacity for a network link is to measure the 5 minute average link load, and then add a safety margin to cater for the effect of burstiness on small time-scales. Because of the substantial measurement efforts required to determine the burstiness, network managers often rely on rules of thumb to find the safety margin, e.g. 'mean plus 50%'. In this paper we propose a novel method to accurately determine the burstiness of traffic on small time-scales, without requiring measurements on such small time-scales. Our method is based on coarse-grained polling of the occupancy of a buffer in front of the link, from which the burstiness on small timescales is inferred. We provide the theoretical foundations of our approach, and a validation through both simulation using synthetic traffic as well as real network traffic taken from various operational networks. It turns out that using our approach, it is possible to accurately determine burstiness on small time-scales (for instance 10 ms), by sampling the buffer occupancy (for instance) every second.

AB - Common practice to determine the required bandwidth capacity for a network link is to measure the 5 minute average link load, and then add a safety margin to cater for the effect of burstiness on small time-scales. Because of the substantial measurement efforts required to determine the burstiness, network managers often rely on rules of thumb to find the safety margin, e.g. 'mean plus 50%'. In this paper we propose a novel method to accurately determine the burstiness of traffic on small time-scales, without requiring measurements on such small time-scales. Our method is based on coarse-grained polling of the occupancy of a buffer in front of the link, from which the burstiness on small timescales is inferred. We provide the theoretical foundations of our approach, and a validation through both simulation using synthetic traffic as well as real network traffic taken from various operational networks. It turns out that using our approach, it is possible to accurately determine burstiness on small time-scales (for instance 10 ms), by sampling the buffer occupancy (for instance) every second.

KW - Network design and capacity planning

KW - queueing network models

KW - IR-63468

KW - METIS-225554

KW - EWI-7069

U2 - 10.1007/11422778_25

DO - 10.1007/11422778_25

M3 - Conference contribution

SN - 978-3-540-25809-4

T3 - Lecture Notes in Computer Science

SP - 303

EP - 315

BT - NETWORKING 2005

A2 - Boutaba, R.

A2 - Almeroth, K.C.

A2 - Puigjaner, R.

A2 - Shen, S.

A2 - Black, J.P.

PB - Springer

CY - Berlin

ER -

Mandjes M, van de Meent R. Inferring traffic burstiness by sampling the buffer occupancy. In Boutaba R, Almeroth KC, Puigjaner R, Shen S, Black JP, editors, NETWORKING 2005: 4th International IFIP-TC6 Networking Conference, Waterloo, Canada, May 2-6, 2005. Proceedings. Berlin: Springer. 2005. p. 303-315. (Lecture Notes in Computer Science). https://doi.org/10.1007/11422778_25