Design And Implementation Of High Performance CDN Streaming Media Server

In recent years,with the diversification of information and the rapid development of network broadband access technologies,audio and video services have become an indispensable element of people's life,making streaming media technology and content distribution network(CDN)more and more popular.At the same time,its related applications such as online live broadcast,small video and other video services are rapidly emerging.Streaming media has the advantages of continuity,realtime,and timing.The combination of streaming media and CDN has become a hot topic developed by major Internet companies.The streaming media server is the core component of data transmission.The traditional streaming media server only relies on the transport layer to control flow which has can no longer meet the needs of large-stream and high-concurrency media servers.The current popular server named Nginx controls the traffic burst by controlling the user's online frequency.But this become more and more limited in practical applications.Therefore,how to control the jitter of traffic and how to optimize the network is a real problem that needs to be solved.In response to these problems,this dissertation takes the actual research project of a large Internet company as the subject.Firstly,after optimizing the CDN architecture we apply the congestion control algorithm based on the traffic shaping strategy to the Nginx server and optimize the Nginx server according to the method of the carrying capacity of adaptive network.Secondly,we modify the token bucket algorithm based on the idea of separating the data cache and the sending module.Finally,we test the system.Our experimental results show that the optimized server avoids transient network congestion effectively caused by data bursts and reduces packet loss greatly compared with the traditional Nginx streaming server.At the same time,it improves traffic control anti-jitter capability,and reduces congestion and jam rate.At last,the user experience satisfaction has been greatly improved which prove the feasibility and effectiveness of the algorithm.