Two-stage replication strategy in P2P VoD systems

In P2P (Peer-to-Peer) VoD (Video-on-Demand) systems, one of the key problems is cache replication. However, with the large scale P2P VoD systems deployed, people study on the log of these systems and gradually realize that the strategy which caches videos proportionally to their popularities is not optimal, especially for cold resources. In simulation, we show that cold videos produce the majority of server workload. In our work, we propose a two-stage cache replication strategy, which increases the distribution of cold resources in the system and reduces the redundant cache of popular resources. Through simulation, we find that the cache proportion of popular videos and cold videos has been readjusted. Owing to more proportion of cache capacity used to cache cold videos, the server workload produced by cold videos is reduced sharply while the server workload produced by popular videos stays low.