More articles in this series and about OTT/Streaming:

As Public Service Broadcasters (PSBs) like the BBC, RAI, ARD, and RTVE move further into being streaming-first, the media distribution landscape is becoming more complex than ever before. In particular, the technical and commercial options facing PSBs for how to stream at the scale and quality they require for their nationwide services are growing, creating more complex choices on how best to architect tomorrow’s streaming solutions.

The PSB Need For Streaming Service Guarantees

Part 1 of this article described the universal service requirement for PSBs and considered how a PSB looks at meeting their challenge of streaming at scale. But while PSBs foresee a future of streaming, the target technical architectures have not yet crystallized. Most PSBs already accept the future is about streaming at scale to very large audiences. This means PSBs require multi-platform service assurance, with guarantees of service delivery quality. But how can service guarantees be achieved in the complex delivery web of the internet? Which organization(s) will provide these guarantees?

Organizationally, PSBs (like all linear TV broadcasters) generally treat streaming as another delivery platform, additive to the other traditional platforms that are currently producing most of their revenue. This view of streaming as “just another delivery platform” makes sense as most of the content produced is the same for all distribution platforms, and content rights deals and advertising deals are worked out by PSB commercial teams across multiple delivery platforms. The primary differences introduced by streaming are for shorter-form content (i.e., for social media), more focus on VOD content (i.e., full series available instantly), and the opportunity for more granular targeted advertising. Streaming really does introduce for PSBs the concept of serving an “audience-of-one”.

PSBs generally have in-house “digital” teams overseeing the streaming application, the website, and the content delivery to these domains. The PSB’s digital Product teams ensure the right features are built into their application to drive up media consumption (video, audio, and text) on the platform, that ultimately is monetized by their commercial teams. Most of the content itself, including linear channel delivery, live event delivery and VOD publication are generally managed centrally because they are multi-platform. Some PSBs are making some content exclusive to their own App to drive up viewership, and most will enable viewing of a complete series on their App even if it is being published on a daily or weekly basis on their linear channels.

Looking into the future, streaming distribution for PSBs is set to become the largest part of their multi-platform delivery model. Pay-TV and Free-to-Air TV Platforms are already aggregating Linear Channels, VOD content, and PSB Apps. And the PSB’s own Direct-to-Consumer (D2C) streaming services make up most of the rest of their streaming distribution. As PSB audiences convert across from traditional broadcast platforms to streaming platforms, the requirement for guaranteed service for streaming increases.

But there is an in-built natural complexity about providing service guarantees for streaming delivery that needs to be overcome as PSBs look at reaching their entire audience.

The Problem Of Streaming Distribution Complexity

The problem with providing a Service Guarantee for Streaming Distribution can be largely attributed to the wide variety of entities in the streaming distribution chain. Other factors that affect this problem relate to the uncertainty of forecasting viewer demand for live events, the uncertainty of managing viewer-level personalization, and the typically multi-purpose use of the networks in the streaming distribution chain.

- Variety Of Entities In Streaming Distribution

On the device-side of the distribution chain one large European PSB estimates they must support over 13,000 discrete Connected TV products. Mobiles, PCs, gaming devices, and set-top-boxes add to the list. And there is an expectation that the device count will only increase.

The networks involved in distributing streaming video then add to this complexity. In a single PSB’s market there are multiple Content Delivery Network (CDN) service providers, multiple Internet Service Providers (ISP) spanning mobile and fixed connections (including the growing number of “AltNets” that are taking market share from larger ISP incumbents), and one or more Access Network Providers. PSBs normally have a single country for distributing their content, but if streaming access is offered to viewers outside of the PSB’s home country then more networks are instantly added to the mix. Much of this complexity is hidden today by CDNs, that handle the underlying network complexity within their operation. If a PSB uses their own CDN infrastructure, this adds extra complexity to manage yet another environment and the underlying distribution networks. If PSBs also use an ISP’s own CDN service in their multi-CDN mix, this also adds to the technical, operational, and commercial complexity of another entity to manage.

Within this network organization set, there are also multiple technologies to use, which adds again to the complexity. Unicast, Multicast, and WebRTC; proprietary CDN technology and Open Caching CDN technology; on-premise, cloud, and edge deployment locations; encoding solutions – the list is long and varied.

Given this overall ecosystem, it is not surprising that streaming distribution is often a best-efforts approach without any guarantees of service quality for a PSB. Each PSB therefore needs to self-insure against performance issues and has an impossible job to definitively guarantee service quality to their whole audience. This is in stark contrast to the reach and service levels PSBs have with traditional broadcast on DTT, DSAT and FM/DAB.

- Uncertainty Of Future Demand

Individual PSBs in countries with populations like the UK, Germany, and Italy will often stream at between 1-5 Tbps per day for normal streaming viewership, while single large events will reach 15-30 Tbps. As broadcast distribution gradually converts to streaming distribution, the forecasts for streaming capacity needs when a full streaming switchover has happened are at least 10 times higher than these figures. Large events will need 150-300 Tbps, in a single country with populations of 50-75 million. This is before we consider immersive viewing experiences that could require a 100 Mbps stream to a device rather than an average of 10 Mbps per device that is assumed in these forecasts. The peak figures could jump up dramatically even if only 5-10% of an audience chooses the immersive viewing option.

Guaranteeing service at the right price is generally founded on guaranteeing capacity and infrastructure over which the service will operate. Handling the Peak to Mean viewing ratio highlighted above presents a problem today for most PSBs, CDNs and ISPs. The difference between Peak and Mean is too big. The future forecasts will make this even worse, as the absolute delta between peak and mean will be orders of magnitude greater than today. But the capacity needs to come from somewhere.

While it would be ideal to base a service guarantee on known capacity requirements, there are two uncertainties for most PSBs about their future demand which makes this complicated. The first is how the daily viewership patterns will evolve given there is generally diminishing viewership of live linear content and a continued migration to time-shifted VOD viewing. Where will this leave the normal daily peaks and means of viewership? The second is how much streaming will really happen on major live events. Will a national event lead to one-time peaks each year/month/week of demand that dramatically exceeds the normal daily peak? Not knowing the answer to these questions leaves PSBs bouncing between wildly different demand scenarios that challenges streaming distribution network planning.

In addition, what will personalized viewing vs. generalized viewing mean? Managing distribution for personalized content needs unicast, while generalized viewing could use multicast. The set-up of the streaming distribution chain is significantly different in each case. So how much content will be delivered in a unicast model vs. multicast model? The crystal-ball gazing continues, and planning network capacities cost-efficiently remains challenging.

- Multi-Purpose Networks

The internet is made up of multi-purpose data networks. This is the beautiful flexibility of IP networks, but also a drawback for PSB-scale streaming distribution.

As mentioned before, guaranteeing service needs to be founded on capacity guarantees. Multi-purpose networks struggle with this. While aggregated content delivery in a multi-tenanted environment can balance out over-forecasting and under-forecasting by individual users of the network, service guarantees need some level of predictability. Predicting that everything is unknown does not help guarantee service quality, and it certainly does not help control costs of service. In an environment like this, the only answer to guarantee service is to dramatically overbuild capacity. But this comes at a high cost. And again, doing this at smaller scales is already complex (witness today’s general lack of streaming service guarantees), but doing it with streaming audiences that are 10+ times larger than today is exponentially more complicated.

The Ideal Way?

With these realities and challenges in front of PSBs and a long list of possible options to resolve them, PSBs are not sure of the ideal way to proceed.

One streaming distribution leader at a large European PSB highlighted for this article that they are not yet seeing a holistic solution for their streaming distribution in future. This person explained that many technical and commercial options are visible, but various pros and cons in each option mean that an ideal choice is not yet presenting itself. They explained that the general view from a PSB perspective today is:

• The future of streaming media distribution will be a multi-platform, multi-technology set-up for two primary reasons. First, consumer devices create options and diversity which must be continuously supported by PSBs to serve their whole audience. Second, network ownership and network technologies are diverse and varied, which PSBs must also support to reach their whole audience. This means that technology and services need to be managed across many domains simultaneously. And this is only for streaming, yet there are traditional existing distribution methods that also need to be managed by the PSB distribution team.
• The gradual maturing of streaming features should mean that centralized functions like encoding, origination, and the media applications themselves should become simpler to manage. Once stabilized, these elements of the media distribution chain should be highly consistent and reliable, like the traditional TVs, Satellite and Cable/IPTV services. The biggest challenge in these areas is likely to revolve around network and platform stability in the compute and storage environments in which these services are running.
• Economies of scale depend on consolidation of suppliers. In a potentially more fragmented network and cloud platform environment, economies of scale could be elusive for a single broadcaster when compared to what multiple PSBs could achieve together. Like broadcasting before it, real economies of scale may only be found by multiple PSBs working together.
• The pioneering pureplay streamers have mostly provided VOD services (e.g., Netflix, YouTube, Disney+), but now major Live events are becoming much more present on streaming-only platforms (e.g., DAZN, Peacock, Amazon Prime, Apple+). These pioneers implemented new streaming platforms, built around their own dedicated and controlled environments, used for the majority of their content delivery. But even with the most controlled environments, there are always Quality of Experience KPIs in the “red” (i.e., not good enough). Streaming is not easy, for all the reasons above, and yet PSBs need to be prepared for even larger audiences than the pureplay streaming pioneers have achieved. If the most advanced streamers today cannot deliver to their whole audience in perfect quality, then PSBs know they must somehow go beyond what even these streamers have been able to do.
• Change is expected to be perennial, like most things involving technology. While we are in a period of transition towards full PSB-scale streaming the level of change in all parts of the ecosystem will be high, particularly the network management aspects that must scale the most. Once full-scale is reached, PSBs should be able to stabilize streaming environments and performance. And in parallel with the change, PSBs and other media businesses have a responsibility to continually improve the efficiency of streaming delivery, in terms of both energy efficiency and capacity-usage. Hybrid solutions mixing multiple delivery technologies (e.g., unicast and multicast) to optimize efficiency across the mix of live and VOD content delivery is expected to become the norm.

What Next?

Each individual PSB needs to manage their own path into a PSB-scale streaming future. But that does not mean that every part of the value chain needs to be managed individually by each PSB in a marketplace. In a market with multiple PSBs, there is an opportunity to collaborate to deliver best practice and cost benefits, particularly in the complex domain of streaming media distribution that uses multiple networks and technology types, and interfaces with a growing number of critical network suppliers.

As the European PSB contributor to this article highlighted, it is time that the PSB Streamer industry could benefit from agreeing what the collective goals are for their streaming media distribution. What those collective PSB goals could be is the subject of Part 3 in this series.