Machine learning based adaptive flow classification for optically interconnected data centers

Nicolaas Viljoen, Houman Rastegarfar, Mingwei Yang, John W Wissinger, Madeleine Glick

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

We optimize flow placement for a hybrid network implementing an adaptive neural network classifier. We predict elephant flows with high accuracy on anonymized university network traffic. We also demonstrate the capability to perform highly complex actions at 40 Gbps using less than 5% of co-processor capacity. This shows that it is possible to implement intelligent actions such as a neural network in a data center using fully programmable NICs without handicapping the server CPU.

Original languageEnglish (US)
Title of host publication2016 18th International Conference on Transparent Optical Networks, ICTON 2016
PublisherIEEE Computer Society
Volume2016-August
ISBN (Electronic)9781509014675
DOIs
StatePublished - Aug 23 2016
Event18th International Conference on Transparent Optical Networks, ICTON 2016 - Trento, Italy
Duration: Jul 10 2016Jul 14 2016

Other

Other18th International Conference on Transparent Optical Networks, ICTON 2016
CountryItaly
CityTrento
Period7/10/167/14/16

Fingerprint

Learning systems
Neural networks
Program processors
Classifiers
Servers
Coprocessor

Keywords

  • circuit-switched
  • networks
  • optical interconnects

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Electrical and Electronic Engineering
  • Electronic, Optical and Magnetic Materials

Cite this

Viljoen, N., Rastegarfar, H., Yang, M., Wissinger, J. W., & Glick, M. (2016). Machine learning based adaptive flow classification for optically interconnected data centers. In 2016 18th International Conference on Transparent Optical Networks, ICTON 2016 (Vol. 2016-August). [7550294] IEEE Computer Society. https://doi.org/10.1109/ICTON.2016.7550294

Machine learning based adaptive flow classification for optically interconnected data centers. / Viljoen, Nicolaas; Rastegarfar, Houman; Yang, Mingwei; Wissinger, John W; Glick, Madeleine.

2016 18th International Conference on Transparent Optical Networks, ICTON 2016. Vol. 2016-August IEEE Computer Society, 2016. 7550294.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Viljoen, N, Rastegarfar, H, Yang, M, Wissinger, JW & Glick, M 2016, Machine learning based adaptive flow classification for optically interconnected data centers. in 2016 18th International Conference on Transparent Optical Networks, ICTON 2016. vol. 2016-August, 7550294, IEEE Computer Society, 18th International Conference on Transparent Optical Networks, ICTON 2016, Trento, Italy, 7/10/16. https://doi.org/10.1109/ICTON.2016.7550294
Viljoen N, Rastegarfar H, Yang M, Wissinger JW, Glick M. Machine learning based adaptive flow classification for optically interconnected data centers. In 2016 18th International Conference on Transparent Optical Networks, ICTON 2016. Vol. 2016-August. IEEE Computer Society. 2016. 7550294 https://doi.org/10.1109/ICTON.2016.7550294
Viljoen, Nicolaas ; Rastegarfar, Houman ; Yang, Mingwei ; Wissinger, John W ; Glick, Madeleine. / Machine learning based adaptive flow classification for optically interconnected data centers. 2016 18th International Conference on Transparent Optical Networks, ICTON 2016. Vol. 2016-August IEEE Computer Society, 2016.
@inproceedings{9218791074a04e06b1b1ed4885eeda6f,
title = "Machine learning based adaptive flow classification for optically interconnected data centers",
abstract = "We optimize flow placement for a hybrid network implementing an adaptive neural network classifier. We predict elephant flows with high accuracy on anonymized university network traffic. We also demonstrate the capability to perform highly complex actions at 40 Gbps using less than 5{\%} of co-processor capacity. This shows that it is possible to implement intelligent actions such as a neural network in a data center using fully programmable NICs without handicapping the server CPU.",
keywords = "circuit-switched, networks, optical interconnects",
author = "Nicolaas Viljoen and Houman Rastegarfar and Mingwei Yang and Wissinger, {John W} and Madeleine Glick",
year = "2016",
month = "8",
day = "23",
doi = "10.1109/ICTON.2016.7550294",
language = "English (US)",
volume = "2016-August",
booktitle = "2016 18th International Conference on Transparent Optical Networks, ICTON 2016",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Machine learning based adaptive flow classification for optically interconnected data centers

AU - Viljoen, Nicolaas

AU - Rastegarfar, Houman

AU - Yang, Mingwei

AU - Wissinger, John W

AU - Glick, Madeleine

PY - 2016/8/23

Y1 - 2016/8/23

N2 - We optimize flow placement for a hybrid network implementing an adaptive neural network classifier. We predict elephant flows with high accuracy on anonymized university network traffic. We also demonstrate the capability to perform highly complex actions at 40 Gbps using less than 5% of co-processor capacity. This shows that it is possible to implement intelligent actions such as a neural network in a data center using fully programmable NICs without handicapping the server CPU.

AB - We optimize flow placement for a hybrid network implementing an adaptive neural network classifier. We predict elephant flows with high accuracy on anonymized university network traffic. We also demonstrate the capability to perform highly complex actions at 40 Gbps using less than 5% of co-processor capacity. This shows that it is possible to implement intelligent actions such as a neural network in a data center using fully programmable NICs without handicapping the server CPU.

KW - circuit-switched

KW - networks

KW - optical interconnects

UR - http://www.scopus.com/inward/record.url?scp=84985914437&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84985914437&partnerID=8YFLogxK

U2 - 10.1109/ICTON.2016.7550294

DO - 10.1109/ICTON.2016.7550294

M3 - Conference contribution

AN - SCOPUS:84985914437

VL - 2016-August

BT - 2016 18th International Conference on Transparent Optical Networks, ICTON 2016

PB - IEEE Computer Society

ER -