Traffic pattern based virtual network embedding
Andreas Blenk, Wolfgang Kellerer · 2013
Virtual Network Embedding (VNE) algorithms are used to find the best embedding of multiple Virtual Networks (VNs) with respect to a performance metric such as network utilization. However, most algorithms assume that the demands of the VNs are static over time. We focus on VNE algorithms for VNs whose demands are changing according to traffic patterns of their provided services. In order to always guarantee high performance and network resource efficiency, the embedding of the VNs has to change, i.e., the VNs have to be re-embedded. By considering traffic patterns of different services, we propose an algorithm that looks ahead and minimize the impact of reconfigurations that are the result of a re-embedding. First simulation results show the potential of VNE algorithms that consider the behavior of evolving service demands.