Abstract

Internet traffic is generated by a multitude of applications, each one with diverse service requirements in terms of bandwidth, latency, reliability, etc. Today traffic engineering techniques can provide service differentiation at the IP/MPLS layer, but not at the optical layer. In this paper we propose a framework where application service requirements drive a dynamic multi-layer (IP/MPLS and optical) resource allocation and optimization. We compare by means of simulations such application-aware algorithmic framework with a multi-layer but application-unaware strategy. Results show that the application-aware approach, unlike the application-unaware one, is always able to guarantee the specified service requirements to those applications whose generated traffic is accepted by the network. In addition, the application-aware strategy does not consume more network resources than the application-unaware one, but only requires a network that is more dynamic and responsive.