Research On Deployment Of Service Function Chain For Cloud Data Center

With the gradual rise of cloud computing and mobile edge computing(MEC)technologies,more and more terminal data floods into the Internet.Traditional network with high coupling and low cohesion architectures are no longer suitable for cloud data centers or edge data center networks.Network service provision ISPs need to converge network function virtualization(NFV)and software-defined networking(SDN)to deploy the service function chains(SFC)flexibly.The ordinary and static SFC mapping mode is too inflexible to promptly perceive the deployment of the virtual network function(VNF)on the physical node,and only a certain dimension of optimization is pursued in the SFC mapping strategy,such as SFC end-to-end delay,mapping cost,etc.In addition,the reusable characteristics of VNF is not considered in most of the existing research on VNF migration.The problem in this paper we mainly study is that the optimization of physical resource allocation required for SFC mapping in data center networks and the multi-dimensional resource load balancing between server nodes.The results of the research are divided into the following two parts:1)For the lack of existing SFC deployment,the SDN controller is used as the control center of the service chain mapping in this paper,and the mapping strategy is dynamically formulated for the incoming service request.First,the KSP algorithm is used to calculate the top K short paths between the SFC source and destination nodes,and then the co-evolutionary multiple population genetic algorithm solves the optimal mapping strategy iteratively.The algorithm ensures the end-to-end delay of SFC and maximizes the resource utilization of physical nodes at the same time,so as to maximizes the resource utilization of the network in a network topology with a limited number of physical nodes and reduces resource fragmentation.2)In view of the shortcoming of VNF migration,the SFC mapping phase combined with the VNF migration phase is proposed in this article to consider VNF migration,to view the VNF migration from the perspective of VNF reusing and establish a reliable VNF migration plan to perform load balancing between nodes.In the SFC mapping phase,the path with the current SFC source and destination with small delay,sufficient average resource is preferentially selected as the mapping path.After the path is selected,the fast-matching algorithm based on the VNF matching degree is used to select the appropriate node on the path as the node to be mapped.VNF matching degree is calculated based on whether VNF multiplexing can be performed.If the resource occupation of a certain dimension of the node reaches the load threshold during the SFC mapping process,the migration of the VNF on the node will be triggered at this time and each VNF which includes the emigration index and the immigration index of the node to be migrated will be calculated according to the characteristics of VNF aggregation and the dimension of load resources.each VNF performs VNF migration based on these two indicators.