Blind data attack on BGP routers

Abstract

Transport Communication Protocol (TCP) implementations may not properly implement blind attack protection, leaving long-standing connections, such as Border Gateway Protocol (BGP) sessions, vulnerable to exploitation. This thesis aims to understand the efficacy of a blind data attack on BGP sessions. This thesis examines BGP, the protocols BGP relies on, and the effectiveness of safeguards against BGP blind attacks. A series of blind attack tests are performed against various production BGP implementations to determine how dangerous and feasible a blind attack is on BGP routing information integrity. Blind data attacks can inject and temporarily propagate erroneous routing information; however, on the routers tested, the complexity required to brute force connection-specific values makes blind data attacks difficult. Also, there is a high probability that a blind data attack will desynchronize a BGP session without modifying routing information. Protective measures are available that could further safeguard BGP sessions, but older router images may not implement some of the most vital protections recommended today. Organizations responsible for routing infrastructure and network security must carefully weigh the risk of not implementing more strict protection measures should a discovered vulnerability reduce attack complexity.