Guaranteeing Availability for Network Function Virtualization with Geographic Redundancy Deployment

Network Function Virtualization (NFV) is a promising technique to greatly improve the effectiveness and flexibility of network services through a process named Service Function Chain (SFC) mapping, with which different network services are deployed over virtualized and shared platforms in data centers. However, such an evolution towards software-defined network functions introduces new challenges to network services which require high availability (HA), especially when catastrophic failures happen, such as earthquake and power outage, which may bring a whole data center down. One effective way of protecting the network services from such failures is to use sufficient redundancy, and in particular geographic redundancy so that when the primary VMs in a failed data center go down, the backup VMs deployed at another location can assume the role of the primary. In doing so, however, the efficiency of physical resources may be greatly decreased. To address such issue, this paper defines availability-aware SFC mapping problem and presents a novel online algorithm that can minimize the physical resources consumption while guaranteeing the required HA within polynomial time. Theoretically, we prove the intractability of the problem and our backup VNFs picking approximation algorithm can achieve a normalized relative error of ((e− 1) ×AE(OPT ) + AE(Ø)) /e, where AE(OPT ) and AE(Ø) are the availabilities with an optimal backup solution and without backups, respectively. Simulation results show that our proposed algorithm can significantly improve SFC request acceptance ratio and reduce resource consumption.