Understanding BGP routing dynamics is critical to the solid growth and maintenance of the Internet routing infrastructure. However, while the most extensive study on BGP dynamics is nearly a decade old, many factors that could affect BGP dynamics have changed considerably. We revisit this important topic in this paper, focusing on not only comparing with the previous results, but also issues not well explored before. We have found that, compared to almost a decade ago, although certain characteristics remain unchanged (such as some temporal properties), BGP dynamics are now 'busier,' and more importantly, now have much less pathological behavior and are 'healthier'; for example, forwarding dynamics are now not only dominant, but also more consistent across different days. Contributions to BGP dynamics by different BGP peers-which are not proportional to the size of a peer's AS-are also more stable, and dynamics due to policy changes or duplicate announcements are usually from specific peers

While the Internet continues to thrive, the resiliency of its fundamental routing infrastructure is not fully understood. In this paper, we analyze the behavior of the de facto inter-domain routing protocol, BGP, during a large-scale power outage that affected the connectivity of 3,175 networks in dozens of cities in the eastern USA and Canada. By observing proper metrics of BGP, we study BGP behavior from both the global level and the prefix level. At the global level, our results show that many global BGP metrics remained stable during the blackout event; importantly, we do not find an increase in the number of BGP announcements, a metric that has been primarily used to indicate significant changes in routing. However, we observe an apparent increase in the number of withdrawals. At the prefix level, we introduce per-prefix AS-path graphs and study their evolution for affected prefixes during the blackout; we have found that in such graphs there is a sharp decrease in the number of edges and nodes as well as changes in node degrees.

Certain popular metrics have been used to reflect the insta- bility of the Internet control plane, such as the volume and duration of BGP updates. However, it is unclear whether there is a direct relation- ship between those metrics and the data plane performance, especially as the Internet is becoming more densely connected and many networks become multi-homed. To clarify this, we measured data streams from a number of PlanetLab nodes toward a sink behind a multi-homed BGP Beacon, which can introduce scheduled BGP routing changes to poten- tially affect the performance of those data streams. In particular, as an important first step, we measured the delay, drop, jitter, and reordering of these data streams and compare them against the volume and dura- tion of BGP updates. We found these data streams were only slightly affected and there is little correlations between these selected metrics from the two planes. Further work includes the correlation of the data plane performance with other control plane metrics while considering more types of routing changes.

Previous studies of the Internet control plane examined its quality from a number of different points of view, but very few considered the performance of the data plane as a measure of control plane performance. For example, a large number of routing updates during some period or slow routing convergence, are often cited as an indication of a poorly designed or behaving routing protocol. However, if the customers’ packets are being delivered well, which we term “happy packets”, while routers can handle routing updates without being overloaded, then how significant are these control plane measurements? We believe packet happiness should be the primary metric for measuring control plane performance, especially the quality of routing. Since handling routing changes is the fundamental function of any routing protocol, in this paper we measure packet happiness during controlled routing changes. We obtain results by using established data plane metrics – delay, drop, jitter and reordering (DDJ&R) – in order to evaluate BGP, the de facto standard inter-domain routing protocol. Moreover, we compare the results with routing updates observed by RouteViews, a commonly used resource for studying the control plane. We establish that generally there is no clear correlation between the number and duration of routing updates derived from partial views and the performance of the data plane, further cautioning the usage of metrics from the control plane alone in evaluating control plane quality.

Abstract: of routing quality9Happy Packets. The measure which counts is whether theusers' packets reach their destination. If the users' packets are happy, the routingsystem, and other components, are doingtheir job. We call these Happy Packets. There are well-known metrics for the dataplane, Delay, Drop, Jitter, and Reordering. So we set out to measure Control Planequality by measuring the Data Plane10Router Scaling. While data plane performance is thegoal, we can't have routers...