Campus Event Calendar
Event Entry
What and Who
Routegazing: Analysing the Evolving Internet Routing Ecosystem
Lars Prehn
Max-Planck-Institut für Informatik - INET
Promotionskolloquium
I'm a last-year Ph.D. student interested in routing, traffic engineering, and congestion control. I work in Anja Feldmann's group at the Max Planck Institute for Informatics in Saarbrücken. My research combines the quantitive analysis of longitudinal data with active probing campaigns to provide new insights into long-standing research topics. A continuous discourse with network operators allows me to contextualize the implications and importance of my findings and guides me towards new, exciting research endeavors.
AG 1, AG 2, AG 3, INET, AG 4, AG 5, D6, SWS, RG1, MMCI
AG Audience
English
Note: We use this to send email in the morning.
Date, Time and Location
Thursday, 6 July 2023
13:00
-- Not specified --
E1 5
002
Saarbrücken
Abstract
The Internet's routing ecosystem constantly evolves to meet the needs of its stakeholders and users. Tracking this evolution is essential, e.g., to identify business opportunities, address security challenges, or inform protocol design. However, most Internet protocols were designed without measurability in mind; hence, many measurements and inference methods rely on exploiting protocol-specific side effects.
This dissertation first assesses the limitations of our deployed observation infrastructures and commonly used inference methods via three orthogonal contributions: a case study on a European Internet Exchange Point to assess our visibility into the Internet's AS topology; a framework to identify and measure biases in the placement of our vantage points across multiple dimensions; and a systematic analysis of the biases and sensitivity of AS relationship inference algorithms. We
found that our view of the Internet's AS topology diminishes over time, and that our AS relationship models are more biased and sensitive to short-term routing dynamics than previously assumed.
With these limitations in mind, we focused on one of the most critical routing ecosystem changes, IPv4 exhaustion, and two ways network operators can deal with it. First, we explored the IPv4 buying and leasing markets, identified market trends, and discussed the viability of these markets for different network types. Second, we analyzed the benefits, usage patterns, and disadvantages of announcing tiny address blocks—which we call "hyper-specific." We argue that a combination of leased IPv4 addresses and hyper-specific prefix announcements likely suffice for many networks to bridge the gap until full IPv6 adoption. Besides its IPv6 adoption, the routing ecosystem also evolved in other dimensions. We first studied AS path prepending to assess the security implication of these changes. We found a typical configuration with no benefits yet an increase of an AS's vulnerability to prefix hijacks. Infrastructural changes led to an overall decrease in prepending sizes
over time and hence a safer use of the technique. However, we demonstrated that we can exploit the same changes to re-orchestrate prefix de-aggregation attacks to overcome widely deployed prevention mechanisms. We validated our assumptions and attack model using a real-world testbed and proposed updates to existing prevention mechanisms. Our two-stage disclosure campaign contributed to a safer routing ecosystem.
This dissertation first assesses the limitations of our deployed observation infrastructures and commonly used inference methods via three orthogonal contributions: a case study on a European Internet Exchange Point to assess our visibility into the Internet's AS topology; a framework to identify and measure biases in the placement of our vantage points across multiple dimensions; and a systematic analysis of the biases and sensitivity of AS relationship inference algorithms. We
found that our view of the Internet's AS topology diminishes over time, and that our AS relationship models are more biased and sensitive to short-term routing dynamics than previously assumed.
With these limitations in mind, we focused on one of the most critical routing ecosystem changes, IPv4 exhaustion, and two ways network operators can deal with it. First, we explored the IPv4 buying and leasing markets, identified market trends, and discussed the viability of these markets for different network types. Second, we analyzed the benefits, usage patterns, and disadvantages of announcing tiny address blocks—which we call "hyper-specific." We argue that a combination of leased IPv4 addresses and hyper-specific prefix announcements likely suffice for many networks to bridge the gap until full IPv6 adoption. Besides its IPv6 adoption, the routing ecosystem also evolved in other dimensions. We first studied AS path prepending to assess the security implication of these changes. We found a typical configuration with no benefits yet an increase of an AS's vulnerability to prefix hijacks. Infrastructural changes led to an overall decrease in prepending sizes
over time and hence a safer use of the technique. However, we demonstrated that we can exploit the same changes to re-orchestrate prefix de-aggregation attacks to overcome widely deployed prevention mechanisms. We validated our assumptions and attack model using a real-world testbed and proposed updates to existing prevention mechanisms. Our two-stage disclosure campaign contributed to a safer routing ecosystem.
Contact
Iris Wagner
+49 681 9325 3500
--email hidden
Virtual Meeting Details
Zoom
637 5637 0417
passcode not visible
logged in users only
Iris Wagner, 06/26/2023 09:25 -- Created document.