basic design with parameters for an entry-level system
basic design with parameters for an entry-level system
This paper proposes a Scalable Architecture for Video on Demand (SAVoD), based on the starting point that it is worth wasting bandwidth to meet latency goals, provided the number of users is high enough to pay for the bandwidth. The design further assumes that eliminating communication with the server is worth a considerable amount of overall bandwidth, as the resulting system can scale to unlimited users who can collectively cover the cost of any wasted bandwidth. A SAVoD entry-level system is proposed, with a relatively small number of movies at relatively low-quality video, which can nonetheless accomodate unlimited numbers of viewers. The entry-level system, once deployed, can scale up to offer close to a full VoD service. Because of the scalability from a small to a large system, SAVoD offers a more practical deployment strategy than other VoD proposals, with fewer compromises than other partial solutions. Furthermore, it is much simpler than other proposed Near Video on Demand (NVoD) solutions, and has the useful property (hence its scalability) of no load or cost impact on the server side or main interconnect for adding another user. Even with a modest 32Kbit/s bandwidth budget per user, a SAVoD system with a large numbers of movies can be supported, with a requirement of 69,000 new users for adding capacity of one new HDTV movie, 23,000 for analog broadcast-quality or 5,800 for an entry-level MPEG-1 system.