Video-on-demand in the Internet has become an immensely popular service in recent years. But due to its high bandwidth requirements and popularity, it is also a costly service to provide. We consider the design and potential benefits of peer-assisted video-on-demand, in which participating peers assist the server in delivering VoD content. The assistance is done in such a way that it provides the same user quality experience as pure client-server distribution. We focus on the single-video approach, whereby a peer only redistributes a video that it is currently watching. Using a nine-month trace from a client-server VoD deployment for MSN Video, we assess what the 95 percentile server bandwidth costs would have been if a peer-assisted employment had been instead used. We show that peer-assistance can dramatically reduce server bandwidth costs, particularly if peers prefetch content when there is spare upload capacity in the system. We consider the impact of peer-assisted VoD on the cross-traffic among ISPs. Although this traffic is significant, if care is taken to localize the P2P traffic within the ISPs, we can eliminate the ISP cross traffic while still achieving important reductions in server bandwidth. We also develop a simple analytical model which captures many of the critical features of peer-assisted VoD, including its operational modes.

Abstract—Gossip protocols are considered very effective to disseminate information in a large scale dynamic distributed system. Their inherent simplicity makes them easy to implement and deploy. However, whereas their probabilistic guarantees are often enough to disseminate data in the context of lowbandwidth applications, they typically do not suffice for highbandwidth content dissemination: missing 1 % is unacceptable for live streaming. In this paper, we show how the combination of two simple mechanisms copes with this seemingly inherent deficiency of gossip: (i) codec, an erasure coding scheme, and (ii) claim 2 °, a content-request scheme that leverages gossip duplication to diversify the retransmission sources of missing information. We show how these mechanisms can effectively complement each other in a new gossip protocol, gossip++, which retains the simplicity of deployment of plain gossip. In a realistic setting with an average bandwidth capability (800 kbps) close to the stream rate (680 kbps) and 1 % message loss, plain gossip can provide at most 99 % of the stream. Using gossip++, ontheother hand, all nodes can view a perfectly clear stream. I.