Research On Optimal Resources Deployment Method Of Service Function Chain In NFV Environment

With the continuous growth of user service demand and the emergence of novel services,operators need to change the current rigid network architecture and implement on-demand provisioning of the substrate network resources.As enabling technologies for future network services,network function virtualization and software defined network can implement dynamic deployment of service function chains by dynamically configuring network functions and performing flexible traffic scheduling between service functions,so as to meet the differentiated performance requirements of various services and applications in terms of latency,reliability and elasticity.Based on the analysis of the current research status of service function chain deployment in the virtualization environment,there are still problems in the following three aspects: 1)the rigid and static service function chain construction method leads to the lack of flexibility in service deployment and it is difficult to perceive the dynamic changes of the substrate network resources,resulting in the in the increase of the service function chain deployment overhead.2)Reactive service function scaling method performs frequent creation and deletion of service function instances,which leads to the decline of service performance and the increase of network operation cost.3)the dedicated or joint backup strategy uses 1:1 resource reservation to improve the reliability of the service,resulting in an increase in resource overhead and service latency.To solve the problems existing in the above research,this paper first uses the double-layers encoding method and the improved genetic particle swarm optimization algorithm to realize the flexible and efficient deployment of service chain.Then,a dynamic deployment method based on resource pre-configuration is proposed to reduce the network operation and service deployment cost.Finally,the resource sharing backup method and hybrid routing strategy are adopted to achieve the guarantee of service reliability and latency.The details of our achievements can be divided into the following three aspects:1.A resource-aware service function chain deployment method is proposed to solve the problem that the static service chain construction method results in the increase of resource overhead and the low acceptance rate of service requests in the scenario of unknown service function sequence.Firstly,according to the dependence between service functions,we design a service function construction method based on the breadth-first search to calculate all service function chain construction schemes corresponding to a service request.Then,with the doublelayers encoding method,the construction and deployment scheme of service chain are encoded in an individual at the same time.Finally,we design a deployment method based on the hybrid particle swarm optimization algorithm is designed to obtain the optimal deployment scheme in accordance with the current network state.The simulation results show that the proposed algorithm reduces the deployment costs of service function chain and improves the acceptance rate of network service requests.2.A dynamic deployment method of service chain based on resource requirement prediction is proposed to solve the problem of service performance degradation and operation cost increase caused by reactive service function scaling method in the scenario of dynamic change of service request arrival rate.Firstly,the GRU neural network is used to predict the service function instance requirements in the future.On this basis,an online service function instance configuration algorithm is proposed to dynamically manage the service function instances to realize the preconfiguration of virtual resources.Finally,a service function path configuration algorithm based on genetic algorithm is proposed to deploy dynamically arrived service requests.The simulation results show that the proposed algorithm can effectively reduce the cost of network operation the occupation of bandwidth resource.3.A reliability guaranteed service function chain deployment method with latency constraint is proposed to solve the problem of resource overhead and service latency caused by the proprietary backup and joint backup methods in the scenario that the substrate network physical nodes may have failure.In this method,a service function sharing backup strategy is proposed to improve the reliability of obtaining network services with less node resource overhead,and then a hybrid routing strategy with single-path and multi-path is adopted to ensure the latency constraint of network services.Finally,the service function deployment problem is formulated as a mixed integer linear programming problem,and a service function chain deployment method based on K-shortest path extension is proposed.The simulation results show that the method reduces the node resources and bandwidth resource overhead of the service chain deployment and improves the deployment success rate.