Font Size: a A A

Optimizing Server Deployment And Routing Strategy In Efficient Content Distribution Network

Posted on:2019-02-08Degree:MasterType:Thesis
Country:ChinaCandidate:H LiaoFull Text:PDF
GTID:2348330569495829Subject:Engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of the Internet,online video has become an important part of the current Internet data traffic.The large amount of video content and the explosive development of video service demand have put the Internet under tremendous pressure.The CDN(Content Distribution Network)has become the main technology for providing such content services.Compared with traditional network architectures,CDN can reduce user access delays,balance service load,reduce network access costs,and improve user's QoE(Quality of Experience).The research of traditional CDN mainly focuses on request redirection,content caching and distribution,and deployment of servers.Many studies have been conducted,but traditional algorithms have been unable to meet the needs of existing networks in the new network environment and new application service demand scenarios(eg,live broadcast,short video).Today's CDN technologies are faced with:(1)In the trend of data center virtualization,when multiICP(Internet Content Provider)share the physical server,resources are wasted due to the unreasonable use of the underlying physical resources;(2)Due to the development of live applications and short video applications,video content is fragmented and content is plentiful and updated rapidly,resulting in frequent redirection of CDNs and frequent updates of content caches;(3)In the mobile Internet scene,the mobility of users brings request dynamics of content,resulting in frequent redirection and frequent migration and updating of content caches.Currently,the deployment and routing control of most CSs(Content Servers)are independently operated and managed by CDNs and ISPs(Internet service providers).Therefore,the deployment of servers affects the traffic demand matrix between network nodes,and it is difficult to maximize the TE(Traffic Engineering)managed by ISPs the role of in content delivery networks,and this problem is more obvious in the multi-ICP scenario.At present,CDN mainly distributes requests to suitable servers for server load balancing through request redirection.There is little research on load balancing for content updates when requested content is missed,and video content fragmentation and user mobility cause frequent updates of content.The current content update is mainly through the latest server or from the SCS(Source Content Server)to obtain updates.When the demand is not balanced,it may cause some servers to be overloaded and thus reduce the service capability of the CDN.To solve the above problem,this paper first considers that the ISP and the CDN share the link state,and considers the user's demand for multiple ICPs to be aggregated in the unit of the cell,and then to minimize the joint deployment according to the network topology and the individual needs of the cell.The cost is the goal of content server deployment,content caching,and joint optimization of transmission routes.On the premise of satisfying the constraints of bandwidth and storage and other resources,a mathematical model was established and a deployment scheme based on hybrid genetic simulated annealing algorithm,pre-screening strategy,and residual bandwidth correction strategy was proposed and simulated.Simulation results show that the deployment scheme can quickly and effectively reduce deployment costs and cache redundancy.Next,due to the fragmentation of content and the dynamicity and effectiveness of various live-like content,it cannot be cached all the time,leading to the problem of unbalanced load that the server frequently updates.We studied the concurrent requests of multiple users in a cooperative caching network.In the case of a miss,consider the goal of load balancing the various update content servers,and then consider how to increase the request acceptance rate while minimizing the cost of updating the route.Under the constraints of server load and link bandwidth,a mathematical model was established.The content server load balancing strategy based on hybrid genetic simulated annealing algorithm and the least cost routing strategy based on greedy policy algorithm were proposed and verified by simulation.The results show that the load balancing deployment solution can effectively control the degree of server load balancing,and after the adoption of the link splitting policy and the same content link aggregation policy,the request acceptance rate and deployment cost have significant benefits.
Keywords/Search Tags:Content Delivery Network, Content Server Deployment, Online Video, Load Balancing, Routing Policy
PDF Full Text Request
Related items