A measurement framework for pin-pointing routing changes

Changes in the end-to-end path between two hosts can lead to sudden changes in the round-trip time and available bandwidth, or even the complete loss of connectivity. Determining the reason for the routing change is crucial for diagnosing and fixing the problem, and for holding a particular domain accountable for the disruption. Active measurement tools like traceroute can infer the current path between two end-points, but not where and why the path changed. Analyzing BGP data from multiple vantage points seems like a promising way to infer the root cause of routing changes. In this paper, we explain the inherent limitations of using BGP data alone and argue for a distributed approach to troubleshooting routing problems. We propose a solution where each AS continuously maintains a view of routing changes in its own network, without requiring additional support from the underlying routers. Then, we describe how to query the measurement servers along the AS-level forwarding path from the source to the destination to uncover the location and the reason for the routing change.