Since the number of Internet users is rapidly increasing day by day and even the most powerful server system will always be resource limited, one of the challenges faced by video-on-demand system designers is how to configure a system that can support a potentially large number of customers and a large multimedia library to satisfy users' needs at affordable rates. In this paper, we propose an approach to provide a significantly scalable video-on-demand service in a multicast environment. The basic idea is to repeatedly transmit popular movies on staggered channels. If a request comes in between staggered start times, the user joins to the most recently started multicast session and then requests the missing part from a nearby neighbor. The user must have enough buffer space to buffer data between staggered transmissions. We refer to our proposed architecture as Neighbors-Buffering Based Video-on-Demand (NBB-VoD) architecture. Analytical results show that our proposal achieves a better performance than the already existing systems (True-VoD, Near-VoD, and Unified-VoD) in terms of both scalability and disk-bandwidth requirements.
- Video-on-demand systems