DSpace@MIT

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 especially useful for fault-tolerant and load-balanced ...