Adaptive Load Sharing in Video-On-Demand Distributed Multimedia System Using Parallel Servers

This Video data is stored in a video server for delivery to multiple receivers over a communications network in a traditional video-on-demand system. The hardware in this video server has the following limitations: 1) the maximum storage capacity, and 2) the maximum number of video sessions that can simultaneously be delivered. We propose a novel Adaptive-Multiple-VIdeo-Server -Architecture (AMVISA) that manipulates parallelism to achieve incremental scalability. This AMVISA architecture 1) exploits data striping in stead of data partition and replication, at the server level to achieve fine-grain load sharing across multiple servers, 2) employs a client-pull service model to eliminate the need for inter -server synchronization, 3) an admission-scheduling algorithm is proposed to further control the dynamic load at each server so that linear scalability can be accomplished. The results in our proposed AMVISA architecture show that the architecture can be linearly scaled up to more concurrent users simply by adding more video-servers and sharing the video data among the servers. The results also support that our new architecture for VoD system is more reliable than single-server VoD system. Key-Words:Video-On-Demand, Video-Server, Distributed Multimedia Systems, Client-Server Architecture, Load Balancing, Load Sharing, Adaptive Resource Management.