What happens when you overshare? A Look into the BGP Maximum-Prefix Feature

The BGP protocol is a crucial component of large network systems, particularly the Internet, as it enables Autonomous Systems (ASes) to exchange information about paths. Although it is essential, BGP is relatively fragile. Sometimes, software errors, misconfigurations, and malicious actors can cause disruptions that range from minor to severe, potentially leading to large-scale outages affecting popular services or companies. Despite efforts to enhance BGP security and configuration, further improvements are necessary in several areas.
One configuration that has received less attention is the Maximum-Prefix Limit, a setting used in BGP sessions to specify the maximum number of IP prefixes that a peer is expected to send. While the intended behavior is for the number of prefixes to stay within this limit, this is not always the case. The two most common ways to handle when the limit is exceeded are: ignore the increase and continue accepting additional prefixes, or temporarily drop the BGP session until the number falls within the limit or a higher value is set. The latter could lead to an outage if many peers are configured to drop sessions.
To assess the likelihood of an outage caused by exceeding the limit, we analyzed 6.5 months of Routing Information Base MRT files from RIPE, from January 1 to July 25, 2025. We also used historical snapshots from PeeringDB to monitor the Maximum-Prefix Limit advertised by ASes.
Our analysis examined how the Prefix Limit, the number of announced prefixes, and the number of peers change over time, while quantifying these changes. Results show that, on average, surpassing the prefix limit does not significantly increase the risk of outages. However, in some instances, it can cause a significant drop in the number of peers, exceeding 80%, which can severely impact connectivity and traffic exchange.