Research On Task Scheduling Technology Of NFV Service Function Chain

NFV technology can dynamically configure Virtual Network functions(VNF)on Service Function Chain(SFC),and flexibly schedule traffic between VNF of different SFC.The service function chain task scheduling service is implemented.However,analyzing the current research status of task scheduling of service function chain in NFV environment,there are shortcomings in the following two aspects:(1)The dynamic selection problem of nodes and the sequencing scheduling problem are transformed into a single static scheduling problem by predetermined fixed parameters,which limits the space of optimization solution,and the traditional heuristic method is easy to fall into local optimum,resulting in the service time extension and instability of the task scheduling of the service function chain.(2)As service requirements change and network request scheduling is phased,network load is unbalanced.A large number of resource fragments reduces the utilization of physical resources,resulting in a low acceptance rate of network requests,wasting physical resources,and increasing energy consumption.In view of the problems in the above research,the specific research content of this thesis is as follows:(1)The SFC task scheduling strategy based on hybrid heuristic algorithm was proposed to solve the problem of extended service time and instability of service function chain task scheduling.Firstly,the network model of VNF scheduling was established,and the goal was to minimize the completion time of service function chain task scheduling.Then,a hybrid optimization algorithm-GABL was designed to optimize the solution,which was composed of artificial bee colony algorithm,genetic algorithm and local search algorithm.The experimental results show that the proposed method has the characteristics of strong stability and fast convergence,and can effectively reduce the network service delay in the task scheduling delay optimization problem of service function chain.(2)A SFC task scheduling strategy based on a Non-dominated Sorting Genetic Algorithm-Ⅲ(NSGAⅢ)with adaptive cross-mutation parameters was proposed,which solved the problems of low utilization rate of physical resources and high energy consumption.The utilization rate of physical resources,network service delay and operation cost are fully considered in SFC task scheduling.Meanwhile,in order to improve the searching and solving ability of the algorithm,Reinforcement Learning(RL)algorithm is used to adjust the crossover and variation parameters of NSGAⅢ algorithm adaptively and keep the diversity of population.The multi-objective optimization strategy is obtained by balancing the global and local search capabilities.Experimental results show that the proposed method can effectively improve the utilization rate of virtual machine resources and reduce task scheduling delay and energy loss.(3)In order to meet the practical application requirements of SFC task scheduling,an easy-to-use human-computer interaction interface is built by using PyQt framework,and an SFC task scheduling system is designed and developed.