For a delay-sensitive application, it is important to maintain the probability that the end-to-end delay is greater than a certain threshold as small as possible. In order to satisfy such a constraint, a shortest path in terms of hop-count is generally chosen. However, hop-count is a very crude measure of the path latency and also if the shortest path breaks before the end of the transmission, it is likely that the delay constraint will not be met. In this paper, we propose an overlay framework that considers both delay and mobility to satisfy the QoS requirements of the delay-sensitive application. Also, we propose network and application layer QoS and mobility management schemes which utilize both reactive and proactive methods.