ElasticSFC: Auto-scaling techniques for elastic service function chaining in network functions virtualization-based clouds

Abstract It is anticipated that future networks support network functions, such as firewalls, load balancers and intrusion prevention systems in a fully automated, flexible, and efficient manner. In cloud computing environments, network functions virtualization (NFV) aims to reduce cost and simplify operations of such network services through the virtualization technologies. To enforce network policies in NFV-based cloud environments, network services are composed of virtualized network functions (VNFs) that are chained together as service function chains (SFCs). All network traffic matching a policy must traverse network functions in the chain in a sequence to comply with it. While SFC has drawn considerable attention, relatively little has been given to dynamic auto-scaling of VNF resources in the service chain. Moreover, most of the existing approaches focus only on allocating computing and network resources to VNFs without considering the quality of service requirements of the service chain such as end-to-end latency. Therefore, in this paper, we define a unified framework for building elastic service chains. We propose a dynamic auto-scaling algorithm called ElasticSFC to minimize the cost while meeting the end-to-end latency of the service chain. The experimental results show that our proposed algorithm can reduce the cost of SFC deployment and SLA violation significantly.