Study On Streaming Data Delivery Based On Hybrid Architecture

With the rapidly development of streaming technology and the higher requirement of users'service quality, how to guaranteeing rapidly and stably transmitting of the streaming files was already an important quality indicator of the streaming technology. For providing higher quantity and more stable streaming media service, satisfying the practical requirements, rigorous demands will be in the transmission of streaming media files from source servers to users, the efficiency of contents distribution and the intelligence of the caching system and so on.In this paper, we introduced peer-to-peer network features , the latest research to streaming based on P2P network streaming system.We introduced streaming media dispersive system model with CDN-P2P hybrid architecture, analyzed the architecture of the system and workflows. To implement the resource sharing of each layers to an extreme, we built up the system with the core of CDN centralized control, proposed two layers P2P network architecture which organizes edge proxy servers and users with the P2P pattern.On the basis above, we researched the two key aspect of streaming media which was the core of this paper. We proposed streaming media segmentation strategy for the streaming files features. Then a pre-distribution strategy for the files data was developed which consists distributed contents, distributed copy numbers and ensuring the position of stored copied. This paper also gave load balancing data distributed strategy in CDN layer including requested data distributing and congestion control. This strategy maintains load balancing of the system to the greatest extent based on the judgment of proxy servers loading situation and the characteristics of streaming media segmentation caching, prevents the absolute imbalance of the system with cyclical congestion control.In the end ,through testing pre-distribution strategy of the streaming media and comparing to the prefix caching pattern, experiments results shows that this algorithm has better performance in reducing users-access delay and promoting efficiency of the servers. In addition, simulation results which based on load balancing streaming data distribution also shows that we made great improve in users-access delay and load balancing.