Conventional SVoD exhibits what I term a high “Crest Factor”. I’ve shamelessly appropriated that term from the telecommunications industry.
In terms of VoD infrastructure, the Crest Factor of content demand determines Server capacity held in reserve, and the buildout of Network infrastructure supporting the peak aggregate bandwidth to all the subscribers.
There are a few dominant factors in VoD crest factor. Time of Day, Day of week, and Blockbuster Content Debut. Demand crests in the evening as subscribers settle down for their evening entertainment. These demand peaks traverse the globe, with the diurnal rhythm of evening twilight. For societal reasons, Sunday night exhibits a more pronounced peak than other days of the week.
Cisco sets the Crest Factor at 5 today rising to over 6 over the next 2 years. This crest behavior continues to grow more rapidly than average Internet traffic. Cisco’s 2019 VNI Forecast (Cisco, 2019) estimates that the “busy hour” Internet use will grow at a CAGR of 37%, compared with 30% percent for average Internet traffic.
What if we could somehow flatten out the Crest Factor issue described above. For argument sake let’s say that that we use the Pareto Principle. In this case 20% of all content represents 80% of all consumption. Let’s not just average out the peaks by time shifting Unicast, let’s further shrink that 80% of the consumption by a much larger factor using Multicast.
The Multicast model in this methodology looks to reduce Infrastructure and bandwidth cost of the Fat Tailed assets by a factor of up to 4,000.
Let’s be conservative. Let’s take only a quarter of that target 4K multiple mentioned above; 1,000 times. Let’s apply this approach to the 20/80 Pareto split; 20% of the media assets consumed by 80% of the subscriber population. 80% of the asset downloads can benefit from a 1000 fold efficiency improvement. 20% of the asset downloads remain unaffected. That’s a cost reduction of >79%. Applied over a multi-year adoption period, Hybrid SVoD could drop capital, and operating expenses to roughly 20% of Conventional SVoD expenses.
The simplistic 1st order nature of these estimates aside, this conjecture represents an impressive Capital and Operational cost saving. Accurate statistics regarding the distribution (shape) of the demand tail is closely held by the Content Service Providers. I suspect that the Pareto distribution is under-stated. Looking at how Netflix has apportioned bandwidth vs storage across its devices on their Content Delivery Network [CDN], it might be closer to a 90/10 spilt. (NetFlix Open Connect, 2019)
Nevertheless, the subscriber count for SVoD is expected to increase by 100% over the next three years, and perhaps double in aggregate video resolution over the next 5. This relief can’t come soon enough to hold the inevitable cost increases down.
Hybrid VoD 