An Empirical Characterization of Anycast Convergence Time

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

78 Downloads (Pure)

Abstract

BGP convergence time is an important aspect of Internet operations and research alike. As an example, traffic steering during DDoS attacks necessitates swift topological changes. Problems surrounding delayed convergence are widely documented. However, little is known about real-world convergence times, especially in an anycast setting. This leads many BGP measurement studies to coarsely estimate the convergence time. In this work, we study BGP convergence using a global anycast testbed. The methodology we present combines two probing methods to measure routing stability from an end-user perspective. We find that 80% of the Internet converges within approximately 10 seconds but also that there is a long tail of networks that are slow to update.

Original languageEnglish
Title of host publicationANRW 2024 - Proceedings of the 2024 Applied Networking Research Workshop
Place of PublicationNew York, NY
PublisherAssociation for Computing Machinery
Pages23-30
Number of pages8
ISBN (Electronic)979-8-4007-0723-0
DOIs
Publication statusPublished - 23 Jul 2024
EventApplied Networking Research Workshop, ANRW 2024 - Vancouver, Canada
Duration: 20 Jul 202426 Jul 2024

Conference

ConferenceApplied Networking Research Workshop, ANRW 2024
Abbreviated titleANRW 2024
Country/TerritoryCanada
CityVancouver
Period20/07/2426/07/24

Fingerprint

Dive into the research topics of 'An Empirical Characterization of Anycast Convergence Time'. Together they form a unique fingerprint.

Cite this