Policy-compliant virtual network embedding

Virtual network embedding algorithms that optimize the mapping of virtual network (VN) topologies onto a substrate network usually do not comply with the policies of substrate providers which may prefer to identify and embed the most profitable subset of a VN request. Such policy-based VN embedding (VNE) is required by distributed VNE architectures, such as PolyVine, or by auction-based VNE environments. In this paper, we introduce a policy dimension to VNE by proposing a new VNE algorithm that aims at maximizing the revenue without violating the provider's policy. In contrast to the greedy nature of most VNE techniques, our algorithm allows a provider to trade short-term revenue gains for higher revenue in the long term and cope better with evolving demands. Our simulation results corroborate the efficiency of our VNE algorithm and show the impact of diverse policy adjustments on resource utilization and generated revenue.