Adaptive service function selection for Network Function Virtualization networking

Abstract

In multiple Service Nodes (SNs) in a Network Function Virtualization (NFV) environment, numerous network functions are carried out. This study concerns the selection of Service Function Instances (SFIs) that are integrated into a Service Function Path (SFP) in resource-constrained networks with various functional requirements. Imbalanced loads in the network system are prevented and a service chain with a high Service Level Agreement (SLA) is developed. An NFV-based Service Function Chain (SFC) integration architecture is designed and an SFP selection mechanism is developed to provide a flexible service chain that satisfies service requests. Two cases are discussed to analyze the effectiveness of the mechanism. The proposed SFP selection mechanism improves the resource utilization of SFIs, prevents the overloading of SFIs, increases the number of service chains that can be serviced by the system, and reduces the packet loss rate. The proposed selection mechanism is compared with without selection mechanism to verify its performance. In case 1, only the load of the SFIs is considered. An SFC controller with the proposed SFP selection mechanism improves resource utilization by 21.32% and reduces the standard deviation of instantaneous loads by 30.02%. In case 2, the acceptable price of instances of service function is considered. The case 2 improves resource utilization by 10.31%, and the standard deviation of the instantaneous load is reduced by 18.73%. An analysis of the rejection rate indicates that the proposed mechanism improves the number of service chains that can be serviced by the system and reduces the mean packet loss rate by 1.91%.