Research On Configuration Problem Of Network Replica Server With Limited Capacity

With the development of Internet and client devices,the number of Internet users in China has also increased rapidly.At the same time,the rapid development of online video,short video,webcast and other industries,especially network streaming websites require a lot of bandwidth.In addition,when remote users frequently visit across regions,the quality of service is significantly reduced.These make the original network system appear more and more weak.Many content providers use content distribution technology to solve these problems,which greatly reduces network traffic,reduces user access delays,reduces the load on the server,and reduces the cost of content distribution,which is also beneficial to web content.Suppliers provide efficient,stable and high quality services.However,in the content distribution network,the way of network replica server deployment and traffic configuration greatly affects network latency,service cost,overall network performance,fault tolerance,and the like when users access content.Therefore,the design of the Replica Configuration Algorithm is very important.This paper studies the capacity-constrained replica server configuration problem on the content distribution network.The main work is as follows:(1)According to the actual problem,this paper proposes a replica server placement model with limited edge capacity based on network flow theory,and analyzes its lower bound algorithm.(2)We studied the static network replica server configuration problem.A set of solution methods is proposed from the feasibility analysis of the problem instance,the replica server location and the network traffic configuration.From the two aspects of mathematical methods and network flow methods,we studied the feasibility analysis of the problem instance.We made use of the optimization method to transform the problem feasibility problem into a linear programming problem,which can be solved within the polynomial time complexity.The problem translates into a maximum flow algorithm that can be solved within the time complexity of O(V E).We studied the network replica server location algorithm,and proposed a greedy algorithm based on network flow.Compared with other location algorithms,our algorithm obtained the best results.We studied the network traffic configuration problem and translated this problem into a minimum cost flow problem.(3)We studied the dynamic network replica server configuration problem and solved the problem of slower solution on the global network.Firstly,the newly added nodes are added into the existing network topology partitioning,and then we proposed a algorithm based on the residual network.After simulation verification,the solution time is reduced by 83.46%.(4)We studied the network topology partitioning algorithm.We proposed a new link weight calculation method based on network flow and obtained the best result.We studied the algorithm of replica server configuration after network topology partitioning.We proposed a new inter-class merging algorithm.(5)We studied the real Internet topology and random network topology generation algorithms,including random geometric graphs,soft random geometric graphs,geographic threshold graphs,Waxman graphs,threshold random geometric graphs and Inet,and implemented the static part of network model simulation platform.We studied the rules and algorithms related to network evolution,and realized the dynamic part of the network model simulation platform.We also studied The method of generating network parameters through real data.