Streaming Media Server System Based On Network Storage

With the development of internet technology and computer software & hardware, multimedia applications have gone deep into our life. Using multimedia technology can receive and play multimedia synchronously, and the technology has been used in video monitoring, web conferencing, video on demand and many other fields. In order to realize those applications above, streaming media server is a key component. The multimedia applications have high requirements to storage space and disk I/O performance especially, while the network storage can meet these requirements well and improve the performance of streaming media server. Therefore, the thesis will research streaming media server system base on network storage under the background of video monitoring.In this thesis, the main work and achievements are as follows:(1) This project bases on an open source streaming media server live555 to realize the streaming media server. The server can connect IP camera with RTSP protocol and realize the function of multi-channel video receiving by multi-process programming and internal process communication. The project has been designed a remote management client by using Socket and MySql database, which improves the ability to manage the receiving status and video files. We test the concurrency performance of streaming media server with different network storage setting, and the result proves the important significance that network storage can improve the performance of streaming media server.(2) The load balancing cluster has been developed base on the single device streaming media server. In the thesis, a new load balancing cluster architecture has been proposed, which remove the load balancer device and distribute the load balancing function to each server in the cluster to raise system reliability. Considering the application background of video monitoring, the thesis proposes a "composite state least connection algorithm" to dispatch task. The utilizing of the load balancing server cluster can improve the performance and the realization of taking over the service of the failed server automatically lead to high reliability of system.