An Algorithm Of VoD Load Balance Based On Service Window

In recent years, media streaming applications over the Internet, such as video conference, IPPhone and Video-on-Demand etc. are getting popular among users. However, the load of media server is too heavy to supply good quality of media data playback due to a number of factors such as limited resources of networks and large numbers of consuming nodes. Hence, how to alleviate effectively workload of media streaming server and assign reasonably resources among different media segments becomes one of primary research.An algorithm of VoD load balance based on service window is proposed in the paper which implements dynamic and proportional assignment of media streaming server resources among media segment. The algorithm organizes the network nodes in P2P way. The system constructs a dynamic tree with fixed height by simplify BMTREE algorithm, insert reasonably new network nodes into the tree through algorithm of nodes management, and implement dynamic adjust among different levels and branches based on status changes of nodes'connectivity during streaming session. The algorithm sufficiently considers status changes of connectivity for each segment during a period of time, proposes notion of weight- connectivity factor to reflect truly popularity degree of each segment for a long time, make that the more popular media segment can obtain more resources to provide connections with better quality for client nodes. In order to answer a paroxysmal request for low popularity degree segment, resource obligator mechanism is intergrated in the algorithm so that a portion of system resources that ensure that system can supplies acceptable server quality can be obtained even though the segment requested have very low popularity degree.Simulations show that the proposed algorithm can assign media streaming servers'resources effectively and proportionally among playback streaming with the resource Reserve, effectively increase the utilization percentage of the system resources.