Dynamic Service-Aware Reservation Framework for Multi-Layer High-Speed Networks

Some current Internet applications (e.g., Grid/Cloud computing storage, video-conference) demand service differentiation, not only in terms of packet forwarding, but also at the connection level. To satisfy the demands for delay-sensitive and low-blocking applications, immediate reservation (IR) and advance reservation (AR) can be performed. Independent resource reservation of each type of network resources is well-investigated. Nevertheless, it is very likely that both types of requests will need to share network resources. There is also significant demand for a service framework that is able to provide application-aware quality of service (QoS). By using existing scheduling algorithms for IR/AR, we develop a service framework to guarantee relative QoS among different application requests, with and without tolerance to delay and service blocking. Simulation results demonstrate the feasibility of our approach and provide a basis for future development of enhanced IR/AR QoS policies within the control plane of future high-speed networks.