By Manu Thapar and Bill Koerner, Hewlett Packard
In the past few years a number of important activities and technology advancements that will influence the future of the industry have emerged. Cable companies have been laying out fiber cable providing enormous amounts of bandwidth capacity to residential neighborhoods. Memory storage is quickly becoming cheaper. Compression and decompression techniques are becoming more well understood and standards such as MPEG are gaining significance. The above factors now enable the possibility of various new services that can be provided to residences. One such service is video on demand (VOD). VOD will allow customers to choose the movies they want to view. Instead of driving to a video store to rent a tape, they will be able to choose a movie that is delivered to their home over a high speed digital network. VOD has the potential of being a multi-billion dollar business in a few years.
The essence of video on demand is that a number of digitized, compressed movies are stored in a video server and transmitted over a distribution network to viewers in their homes.
The overall architecture is shown in figure 1.
There are a number of possibilities for the distribution network [ 1]. The bandwidth into the residence is limited by the final interface at the home. Due to cost constraints, the equipment in the home is assumed to be minimal, comprising little more than decompression, decoding, demodulation, and display on a TV. The movie is played back in real time over the network and is not downloaded into storage in the home. The system cost is dominated by the cost of storage, and it is less expensive to concentrate the storage in the server where only enough storage for the number of simultaneous viewers is needed, rather than having idle storage in the homes of all non-viewing subscribers.
This paper describes the key issues in the design of video servers. We provide an overview of some of the video server architectures currently being explored. We also briefly overview the potential services, service requirements, features, and communication facilities required for the whole infrastructure. We highlight the design issues that to us are the interesting research problems and outline some of our views on their solutions.