| Network Function Virtualization(NFV),as an emerging technology,decouples network functions from dedicated hardware and implements in the form of software on commodity servers through virtualization and cloud technologies,which is called Virtual Network Function(VNF).Therefore,compared with traditional networks,the emergence of NFV makes network service deployment more flexible and agile,bringing great scalability to the network.In the NFV-enabled network,the data flow of network service needs to follow a strict sequence through a series of virtual network functions.The ordered set of VNFs is also called the Service Function Chain(SFC).To ensure that more SFC requests are supported for users in different environments,operators need to design an effective method to reasonably deploy VNFs to appropriate locations considering resource constraints on physical nodes and links.This process is called SFC initial mapping optimization.In addition,during the operation phase of network functions,user service traffic is constantly changing,which may lead to network node and link overloading,and user end-to-end delay conflicts.In this case,it is necessary to reschedule the running SFC and migrate the related VNFs to ensure that the network has good scalability.This process is called SFC dynamic reconfiguration optimization.In recent years,the research on the mapping and reconfiguration mechanism of SFC has attracted extensive attention from the industry and academia,and there have been some excellent works.But there are still some problems to be solved.First,in the design of the SFC mapping mechanism,for effectively balancing the resources of network links and the utilization of the VNF,and the existing SFC mapping optimization method still has some space to be improved.Secondly,when network nodes are overloaded or user Quality of Service(QoS)is degraded,a reasonable SFC migration scheduling scheme needs to be designed to take into account the performance among network load balance,migration cost and service delay.Therefore,in view of the above problems,this paper studies the mapping and migration optimization mechanism of SFC under different network environments,and makes the following contributions:1.In the SFC mapping process for data center networks,to solve the conflict between link occupation and VNF utilization effectively,the SFC mapping problem are formulated as an integer linear programming model which considers minimizing the the link occupancy in SFC mapping problem while ensuring an effective SFC acceptance as the optimization goal.Then,we propose the VNF correlation determination method based on self-learning matrix.Based on this method,the correlation-aware VNF mapping algorithm is designed to solve this problem.The algorithm aims to solve the contradiction between link occupation and VNF utilization through the correlation aggregation of VNFs.Finally,we conduct a large number of evaluation experiments under the data center network architecture.Experimental results show that our approach saves about 17%～54%bandwidth resource,and has better performance in SFC acceptance and VNF utilization rate.2.In the of SFC reconfiguration process for data center networks,we assume that each deployed VNF is used by multiple SFCs and deal with the optimal location allocation for the concurrent migration of VNF s based on the actual network situation.We formulates this problem as an integer programming model which aims to minimize the end-to-end delay for all affected services and to guarantee network load balancing after the migration simultaneously.To this end,we propose the improved hybrid genetic evolution algorithm to address it.In addition,to reduce the computation overhead for large-scale networks,a multi-stage heuristic algorithm based on optimal order is also designed.Finally,extensive evaluation shows that compared with the previous algorithms,the proposed approaches can effectively reduce the SFC average delay by about 11%～24%for different scale networks while ensuring network load balancing.3.In the of SFC reconfiguration process for mobile edge networks,users are constantly moving and different users usually have different delay requirements for service requests.This paper focuses on the SFCs reconfiguration scheme with resource capacity constraints in the MEC network to support the seamless migration of mobile user services.We first formalize the SFCs reconfiguration problem of the edge network as an integer programming model,which aims to minimize the end-to-end delay and operating costs of user services.To achieve the seamless migration of SFC,we propose the migration algorithm based an Dijkstra to obtain the set of migration paths.Then,we convert this problem into an equivalent shortest path problem and design a Dynamic Programming-based SFC Migration algorithm(DPSM).Finally,we conduct simulation experiments to evaluate the performance of the DPSM algorithm based on a real-world dataset.The experimental results show that our approach is better than previous studies,and can effectively reduce the migration cost of operators by about 21%～54%while ensuring user service delay. |
