Fine-grained failover using connection migration

This paper presents a set of techniques for providing fine-grained failover of long-running connections across a distributed collection of replica servers, and is espe-cially useful for fault-tolerant and load-balanced delivery of streaming media and telephony sessions. Our system achieves connection-level failover across both local- and wide-area server replication, without requiring a front-end transport- or application-layer switch. Our approach uses recently proposed end-to-end “connection migra-tion ” mechanisms for transport protocols such as TCP, combined with a soft-state session synchronization pro-tocol between replica servers. The end result is a robust, fast, and fine-grained connec-tion failover mechanism that is transparent to client ap-plications, and largely transparent to the server applica-tions. We describe the details of our design and Linux implementation, as well as experimental data that sug-gests this approach is an attractive way to engineer ro-bust systems for distributing long-running streams; con-nections suffer relatively small performance degradation even when migration occurs every few seconds, and the associated server overhead is small.