Cost-Efficient Service Assurance with Joint Orchestration in NFV Networks

Network Function Virtualization (NFV), as an enabler for 5G and network slicing, introduces new abstraction levels and orchestration to automate the network operations. However, such features make it challenging to assure the performance of network services provisioned with Virtual Network Functions (VNFs) timely and accurately. In this paper, we propose a joint service assurance and orchestration approach to assure the End-to-End (E2E) service in a closed loop. The joint approach integrates performance monitoring, Load Balancing (LB), data analytics, and NFV orchestration. To minimize monitoring cost, a selection discipline is designed to monitor only a subset of E2E flows. Based on a Minimum Deviation (MD) metric, the measured E2E delays are examined to detect Service Level Agreement (SLA) violation and identify the problematic VNF instance(s). The MD metrics are also used in LB that continuously adjusts the traffic load assigned to each flow as a proactive way of preventing delay degradation as early as possible. Furthermore, now that the faulty VNF instances are identified, the orchestrator can take corrective actions such as VNF migration to restore the E2E service performance as quickly as possible. A series of simulations are run to verify the efficiency of the proposed joint approach.