Cost-aware service function chain orchestration across multiple data centers

Network function virtualization is a new network architecture, where the dedicated hardware network functions can be implemented in network function instances running on general purpose hardware such as high volume servers in data centers. End-to-end services require the traffic flow go through a list of NFs in sequence, which is defined by service function chain (SFC). Multiple NFs in a SFC are often orchestrated across multiple DCs to satisfy their position or performance requirements. However, different orchestration strategies of the SFC will lead to different deployment cost, including VNF instance cost and inter-DC bandwidth cost. Besides, large number of NFV instances are deployed in micro-DCs which have limited physical resource. Therefore, in this paper we investigate a costaware strategy to orchestrate the SFCs across multiple DCs, while considering the loads of DCs. An Integer Linear Programming (ILP) model is formulated to minimize the total deployment cost. Then, we prove that the problem is NP-hard and provide a heuristic Cost-Aware SFC Orchestration algorithm (CASO) to solve it. The simulation results show that CASO orchestrates SFCs in a cost-efficient way.