Zeph & Iris map the internet: A resilient reinforcement learning approach to distributed IP route tracing - LAAS-Réseaux et Communications Accéder directement au contenu
Article Dans Une Revue Computer Communication Review Année : 2022

Zeph & Iris map the internet: A resilient reinforcement learning approach to distributed IP route tracing

Résumé

We describe a new system for distributed tracing at the IP level of the routes that packets take through the IPv4 internet. Our Zeph algorithm coordinates route tracing efforts across agents at multiple vantage points, assigning to each agent a number of /24 destination prefixes in proportion to its probing budget and chosen according to a reinforcement learning heuristic that aims to maximize the number of multipath links discovered. Zeph runs on top of Iris, our open-source system for orchestrating internet measurements across distributed agents of heterogeneous probing capacities. We show that carefully choosing which destination prefixes to probe from which vantage point matters for optimizing topology discovery and that a system can learn to improve its assignments based on previous measurements. After 10 cycles of probing, Zeph is capable of discovering 3.3M nodes and 19.8M links in a cycle of 15 hours, when deployed on 5 Iris agents. This is 3 times more nodes and 10 times more links than the existing state-of-the-art production system for the same number of prefixes probed.
Fichier principal
Vignette du fichier
Zeph___Iris_CCR_2022_paper___camera_ready-10.pdf (663.02 Ko) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-03597580 , version 1 (04-03-2022)

Identifiants

Citer

Matthieu Gouel, Kevin Vermeulen, Maxime Mouchet, Justin Rohrer, Olivier Fourmaux, et al.. Zeph & Iris map the internet: A resilient reinforcement learning approach to distributed IP route tracing. Computer Communication Review, 2022, 52 (1), pp.2-9. ⟨10.1145/3523230.3523232⟩. ⟨hal-03597580⟩
149 Consultations
299 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More