Dynamic Fault Management in Service Function Chaining

Service Function Chaining (SFC) is an important mechanism in Software-Defined Network/Network Function Virtualization (SDN/NFV) to allow the network traffic to be forwarded through a pre-defined ordered set of Virtual Network Functions (VNFs). Nevertheless, such a chain may be broken with the failure of a VNF. In this research, we propose a novel fault management system that can rapidly and dynamically recover an SFC from a failed VNF. In our proposed system, no backup VNF or backup Service Function Path needs to be established in advance. Our system will monitor all the VNFs used by SFC and when there is a VNF out of service, it will dynamically create a new VNF with the same functionality as that of the failed one and replace it. Unlike existing solutions of fault-recovery using virtual machines, our system employs container technologies to speed up the instantiation and replacement of VNFs. When compared to the existing method, our proposed solution is capable of achieving much faster recovery time with acceptable cost of resource consumption.