Localizing link failures in legacy and SDN networks

Akbari Indra Basuki, Fernando Kuipers

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review

2 Citations (Scopus)
221 Downloads (Pure)

Abstract

Localizing network link failures is crucial to guarantee sufficient network capacity and to efficiently manage network resources. However, since most of today's networks use link aggregation to increase bandwidth, localizing a single physical link failure within such aggregated links is challenging. In this paper, we propose and evaluate methods, for both legacy networks as well as SDN networks, to localize link failures in the presence of aggregated links.For legacy networks, we propose STreLo for localizing single link failures. We compare STreLo to a state-of-the-art solution, called SCMon, and show that, at the expense of using more probe packets, it is faster and uses less MPLS labels. Since probe packets are small, we deem the trade-off beneficial.For SDN networks, we propose StaF, which works in a decentralized way, requires no controller interaction, and can adapt to topological changes. Moreover, StaF can localize multi-link failures. Both approaches have been tested via Mininet implementations and experiments.

Original languageEnglish
Title of host publicationProceedings of 2018 10th International Workshop on Resilient Networks Design and Modeling, RNDM 2018
PublisherIEEE
Pages1-6
Number of pages6
ISBN (Electronic)978-1-5386-7030-9
ISBN (Print)978-1-5386-7031-6
DOIs
Publication statusPublished - 2018
Event10th International Workshop on Resilient Networks Design and Modeling, RNDM 2018 - Longyearbyen, Svalbard, Norway
Duration: 27 Aug 201829 Aug 2018

Conference

Conference10th International Workshop on Resilient Networks Design and Modeling, RNDM 2018
Country/TerritoryNorway
CityLongyearbyen, Svalbard
Period27/08/1829/08/18

Fingerprint

Dive into the research topics of 'Localizing link failures in legacy and SDN networks'. Together they form a unique fingerprint.

Cite this