Efficient Joint Approaches for Location-Constrained Survivable Virtual Network Embedding

With the development of datacenter networks and network virtualization, facility-failure-proof survivable virtual network embedding (SVNE) has recently gained notable attention. In this work, we design novel location-constrained SVNE (LC-SVNE) approaches that consider the working and backup embeddings jointly and protect virtual networks against single-facility-failures efficiently. We first transform the survivable node mapping to a bipartite graph matching problem and solve it for effective resource sharing among working and backup facilities. Then, two survivable link mapping schemes are designed to minimize the bandwidth consumption of virtual links, by leveraging anycast- and multicast-based substrate routing scenarios. Simulation results show that the proposed approaches outperform two existing ones on blocking probability and time-average revenue.