Optimized Deployment Of CDN In NFV Environment

Migrating traditional content distribution networks to the environment of network functions Virtualization can bring many benefits,such as easy management and low deployment costs and so on.In addition,the network function virtualization technology makes it possible to make a pay-as-you-called pricing model.So how the virtual content delivery network reduces the use of resources to reduce operating costs becomes an important issue.In addition,the traditional content delivery network is a typical representative of large-scale distributed network,and the system size is based on peak traffic.In the non-peak case,the traditional content delivery network energy efficiency is low and energy consumption is large.In the network function virtual environment,the system size adjustment is more convenient than the traditional environment.So we can dynamically adjust the size of its system to achieve the purpose of energy conservation.So this article discusses above issues.Because the running cost of the virtual content delivery network is directly related to the location of the replica functions.So in the second chapter of this article,the replica functions placement problem in vCDN is studied.Thesis will deploy the content delivery network replica function in the virtual underlying network to reduce the operating costs of the virtual content delivery network.Due to the operating cost of vCDN system is mainly determined by the locatin of the virtual function,this chapter givers the virtual function placement algorithnm based on the spectral clustering.This approximation algorithm named SC_CDN(Spectral Clustering Best Function Placement Algorithms for CDN).In the SC_CDN algorithm,we use the idea of spectral clusreing algorihtmn to transform the connection relation between the virtual underlying netowork nodes into the similarity matrix.And then use the spectral clustering algorithm to sub-map the underlying topology to determine the scope of each copy function.Finally enumerate the nodes in each subgraph to find the optimal placement.This can effectively redece the virtual content delivery network operationg costs.In the third chapter of this article,the thesis will study the virtual content delivery network energy saving problem.Though dynamically adjust the virtual content delivery network system size,the algorithm in this thesis can improve the virtual content delivery network energy efficiency to save energy.In the NFV environment,the virtual content delivery network is measured by software components deployed on a common device.So the operation of vCDN will be more flexible and reliable compared with the traditional CDN.It maybe a practical solution by dynamically adjusting the size of the system to adapt to changes in traffic conditions to save energy.And thesisdesigns an energy saving scheme based on time series prediction algorithm MFF(the vCDN energy save algorithm based on max flow sequence forecasting).In the MFF algorithm,we count the maximum value of the flow in each time period and then use the linear time series prediction model ARIMA to predict it.MFF plans the number of system functions required for the next time period based on the traffic forecast value.In addition,the MFF algorithm also designed redundant capacity to improve the quality of the vCDN system.Through the above program,MFF can effectively reduce the vCDN energy consumption.