On design of network scheduling in parallel video-on-demand systems

Introduction. Due to the fundamental limitation of the singleserver architecture, it suffers from the problems of scalability and server fault tolerance. To effectively address these two issues, designing parallel video servers become inevitable [l]. Because of the nature of timely video data, end-to-end resources in parallel video servers, such as disks, CPU and network, have to be reserved and properly scheduled in order to guarantee continuous hiccup-free display in clients. Many research works have already demonstrated the impediment of network bandwidth in the endto-end path. In this research, sufficient disk bandwidth without causing the bottleneck is assumed and the contribution mainly focuses on exploring scheduling algorithms to optimize the utilization rate of network bandwidth in shared and switching interconnection network environments for Constant Bit Rate (CBR) and Variable Bit Rate (VBR) video streams,