Application-Driven TCP Recovery and Non-Stop BGP
Some network protocols tie application state to underlying TCP connections, leading to unacceptable service outages when an endpoint loses TCP state during fail-over or migration. For example, BGP ties forwarding tables to its control plane connections so that the failure of a BGP endpoint can lead to widespread routing disruption, even if it recovers all of its state but what was encapsulated by its TCP implementation. Although techniques exist for recovering TCP state transparently, they make assumptions that do not hold for applications such as BGP. The authors introduce application-driven TCP recovery, a technique that separates application recovery from TCP recovery. They evaluate their prototype, TCPR, and show that it outperforms existing BGP recovery techniques.