Cyclops: The AS-level Connectivity Observatory

By: 
Ying-Ju Chi, Ricardo Oliveira, and Lixia Zhang
Appears in: 
CCR October 2008

In this paper we present Cyclops, a system that collects and displays information of AS-level connectivity extracted from looking glasses, route-servers and BGP tables and updates of hundreds of routers across the Internet. From an operational standpoint, Cyclops provides ISPs a view of how their connectivity is perceived from the outside, enabling a comparison between their observed connectivity and their intended connectivity. ISPs can use the tool to detect and diagnose BGP misconfigurations, route leakages or hijacks based on false AS path. From a research standpoint, Cyclops is able to provide a quick snapshot of the AS-level connectivity of each network, and provides mechanisms to infer the root cause of BGP (de)peerings, which is an important input to the study of AS-level topology models and inter-domain routing protocols.

Public Review By: 
Dmitri Krioukov

Can you hear the shape of the Internet? If not, can you see it? Perhaps you would be better off if you are a cyclops. Focusing your single eye on a single AS, you would be able to get a clear view of the AS Internet structure and dynamics at this AS point ("pupil") surrounded by incident AS links ("iris"). Of course, you can refocus, too, but not too fast, or you will get dizzy.
This is the idea behind the Internet AS topology visualization tool presented in the following paper. The authors call it Cyclops (http://cyclops.cs.ucla.edu/). They collect AS topology data from a variety of sources (RouteViews, RIPE, route servers, etc.) and provide a per-AS interface to it. The user selects an AS and then can check the connectivity of this AS via either the visualizer or the Web interface showing tables with rows of annotation information per incident AS link. The raw data is also available to use in custom scripts. The AS links are annotated with AS relationships (customer-provider or peerpeer), types of neighboring ASs (Tier-1, large/small ISP, stub, etc.), their degrees, times of their first appearance or disappearance, and other information. Perhaps the most useful feature is that the user can track the changes in the connectivity of a given AS over a specified time period. The paper concludes with some case studies (AS route leakage, outage, and depeerings) that utilize this feature.
Overall, the paper presents a nice interface/visualization tool for Internet AS topology studies that the Internet research community and ISPs may find useful.