Thesis: Detecting Hidden Broken Pieces of The Internet: BGP Lies, Forwarding Detours and Failed IXPs

PhD defense: Julian Martin DEL FIORE

Title: Detecting Hidden Broken Pieces of The Internet: BGP Lies, Forwarding Detours and Failed IXPs

Team: Réseaux

Abstract: The Internet is simply an interconnection of independent networks, known as Autonomous Systems (ASes). Given that ASes are built on top of hardware and software, and that network operators, i.e., humans, manage ASes, then the Internet is constrained to some limitations. For example, humans are error-prone and eventually take arbitrary decisions, enterprises are generally greedy from a revenue point of view, and hardware may fail and require maintenance or replacement. All these factors may lead the Internet to have broken pieces, i.e., malfunctioning components, networks facing limitations and even selfish networks prioritizing their own revenue rather than the better performance of the Internet.
The objective of this thesis is to detect broken pieces of the Internet. First, we study the deployment of Internet exchange points (IXPs) in Latin America, a region that has previously received little attention in Internet studies. We construct the most comprehensive dataset of the status of the Internet in Latin America and characterize the AS ecosystem in the region. We find that while some IXPs across Latin America have managed to proliferate, some countries have failed IXPs, i.e., no IXP at all, or the IXP has not succeeded to attract members. Second, we study whether ASes carry on BGP lies, i.e., if the forwarding routes through which packets actually flow on the Internet diverge from the AS-paths that ASes advertise on BGP, the routing protocol used on the Internet. In practice, performing this comparison is complex since besides the multiple levels at which data needs to be synchronized, missing hops, third-party addresses and AS siblings may introduce errors by wrongly triggering the detection of BGP lies. In particular we develop a methodology allowing to filter this noise, and run measurements in the wild. We find cases where after sanitizing the dataset with our framework, paths still mismatch. Finally, we study how traffic flows inside ASes and focus on the detection of forwarding detours, i.e., cases in which the forwarding routes do not match the best available routes, according to the routing protocol in use. We develop a formalism explaining when forwarding detours occur, and implement a detector allowing to differentiate forwarding detours from load balancing and traffic engineering techniques. We run measurements with our detector and find detours in multiple ASes with a remarkable binary pattern such that transit traffic traversing between two border routers of an AS either never detours, or always does.

This defense will be held in English and will take place on 8th Feb 2021 at 1:00pm. This defense will take place in the presence of the jury members listed below:

• Cristel PELSSER, ICube, Université de Strasbourg, France (Directeur de thèse)
• Pascal MERINDOL, ICube, Université de Strasbourg, France (Co-encadrant)
• Benoit DONNET, Department of Electrical Engineering and Computer Science Montefiore Institute, University of Liège, Belgium (Rapporteurs)
• Guillaume URVOY-KELLER, I3S Research Laboratory, University of Nice Sophia Antipolis, France (Rapporteurs)
• Clémence MAGNIEN, LIP6 laboratory, Sorbonne Université, France (Examinateurs)