Skip to Main Content
The Border Gateway Protocol (BGP) is the de facto interdomain routing protocol on the Internet which controls the packet forwarding behavior on the data plane. It has significant impact on the well-being of the global Internet. Over the past ten years, there has been a large body of studies conducted on evaluating and improving the BGP performance. These studies develop tools using BGP data for identifying the Internet topology, AS relationships, and AS-level paths. More importantly BGP is the main data source for evaluating the Inter-domain routing performance and discovering routing anomalies such as prefix hijacking attacks. However, most of these studies focus on one or a few aspects of BGP in a short time period. Till today, the route monitoring system has been deployed for ten years and there has been a significant amount of criticisms on the bad performance of BGP. Our work is the first to critically examine and summarize BGP performance and its changes through time. We evaluate BGP from a diverse set of aspects ranging from routing diversity to convergence performance. We design a set of systematic statistical analysis to cope with the noise in data collection process. Due to the huge volume of data required for the analysis, we implement our evaluation system on top of the cloud computing platform from Amazon EC2. Our results provide a few insights on how to improve BGP and the Internet routing system in the future.