By Winston Hodge, Hodge Computer Research and Chuck Milligan, Storage Technology Corporation
Video On Demand (VOD) has the potentialof giving individual television viewersnearly instant access to a wide range ofrecorded movies, video programs, games,information and other services. It is distinguishedfrom more conventional TV viewingby a high degree of interactivity between theviewer and the material being viewed.
A perception exists in this industry todaythat each person interacting with their TVdemands instantaneous response. This iscalled True Video On Demand (TVOD). Asthis paper will show, TVOD is extremelyexpensive when it provides for all servicespossible.
The alternative to TVOD is Near VideoOn Demand (NVOD). This paper willdemonstrate that while NVOD is significantlyless expensive to implement, an NVODsystem can be designed so that its delays arenot objectionable to the user for manyapplications. Procedures and strategies forconcealing customer latency time will bedescribed, along with the cost differentialattendant to eliminating it.
Access to recorded material with zeroaccess time is not physically possible.
Fractional second access is possible, butwould be very expensive for an unlimitedmenu of choices by an unlimited number ofsubscribers.
Clearly, the quantification of cost to provideservice versus the latency time is ofserious importance. But there is more to theimplementation decision than cost. Thepsychological effects of waiting come intoplay. For example, is one second too long towait? How about two seconds? How abouttwo minutes? All things being equal, (whichthey are not), the shorter the service time thebetter.
This paper will provide a clear view ofphysically possible service times and the costto provide those services using advancedtechnology hierarchical storage.
A model will be described which demonstrateshow the system cost varies withviewer latency. This model will be appliedseparately and collectively to the videoserver, disk storage complex, large terabyterobotic tape farms, VOD selector switch,communications channel and viewer selectionmechanism.
Block diagrams used in the systemsanalysis and simulation will be included,along with charts and graphs which willclarify the results of the analysis. The paperwill conclude with recommendations for aneconomically viable system design.