Dynamic VNF Scheduling for Network Utility Maximization

In this chapter, a delay-aware VNF scheduling problem is studied in the presence of small-timescale traffic dynamics, to achieve network utility maximization with QoS guarantee for each deadline-constrained service. A state-of-the-art time quantum in the 100 μs to ms granularity is used for CPU resource scheduling. Based on the Lyapunov optimization technique, an online distributed VNF scheduling algorithm is derived, which greedily schedules a VNF at each NFV node based on a weight incorporating the backpressure-based weighted differential backlogs, the throughput performance, and the packet delay. Simulation results demonstrate an utility-backlog trade-off with utility importance parameter. The delay-aware VNF scheduling algorithm using a realistic time quantum achieves a comparable performance with a generalized processor sharing (GPS) scheme under the assumption of infinitely divisible resources.