What is the next hop to more granular routing models?
Abstract
Despite its widespread use, the “Gao-Rexford” model has long been recognized for its limitations in accurately capturing Internet routing behavior. However, the root causes of these limitations remain poorly understood, due to the lack of ground truth data. We address this by systematically analyzing inference techniques against generated topologies. Our findings reveal that the greatest issue with existing models is their lack of granularity, rather than the lack of data used to infer them. To overcome this limitation, we extend model granularity by incorporating internal topologies (at the router level) along with broader interdomain and intradomain routing policies. Additionally, we introduce an efficient pathfinding algorithm capable of computing router-level paths on Internet-scale topologies. Our results demonstrate that our extended model significantly improves path accuracy, reaching 86% with only 5% of vantage points—up from the 57% achieved by existing techniques with full visibility. These findings highlight the potential of finer-grained models to enhance path prediction and set the stage for a promising research agenda.
People
BibTex
@INPROCEEDINGS{kirci2024granular,
isbn = {979-8-4007-1272-2},
doi = {10.1145/3696348.3696859},
year = {2024},
booktitle = {HotNets '24: Proceedings of the 23rd ACM Workshop on Hot Topics in Networks},
type = {Conference Paper},
author = {Kirci, Ege Cem and Torsiello, Valerio and Vanbever, Laurent},
abstract = {Despite its widespread use, the "Gao-Rexford" model has long been recognized for its limitations in accurately capturing Internet routing behavior. However, the root causes of these limitations remain poorly understood, due to the lack of ground truth data. We address this by systematically analyzing inference techniques against generated topologies.Our findings reveal that the greatest issue with existing models is their lack of granularity, rather than the lack of data used to infer them. To overcome this limitation, we extend model granularity by incorporating internal topologies (at the router level) along with broader interdomain and intradomain routing policies. Additionally, we introduce an efficient pathfinding algorithm capable of computing router-level paths on Internet-scale topologies.Our results demonstrate that our extended model significantly improves path accuracy, reaching 86% with only 5% of vantage points---up from the 57% achieved by existing techniques with full visibility. These findings highlight the potential of finer-grained models to enhance path prediction and set the stage for a promising research agenda.},
keywords = {Internet routing; BGP; Topology inference; Path computation},
language = {en},
address = {New York, NY},
publisher = {Association for Computing Machinery},
title = {What is the next hop to more granular routing models?},
PAGES = {343 - 351},
Note = {23rd ACM Workshop on Hot Topics in Networks (HotNets 2024); Conference Location: Irvine, CA, USA; Conference Date: November 18-19, 2024}
}Research Collection: 20.500.11850/695737