Fast optimal video placement algorithms for hierarchical video-on-demand systems

The multilevel hierarchical network architecture has been shown to be a scalable and cost efficient solution for large video-on-demand (VOD) systems. The predominant operation cost of a hierarchical VOD system consists of network transmission cost and video storage cost. How to minimize the operation cost under several operating constraints is an important issue. Many operating constraints, such as the storage capacity limitation at each level of servers, have made the problem intractable. We proposed several efficient heuristic video placement algorithms that can achieve near optimal operating cost. We have also proposed a time-variant arrival traffic model with arrival rate matching the statistics gathered from commercial systems.