Hopes and challenges for cloud-hosted broadcasting

With a few months until kick-off, this year's FIFA world cup is already proving to be a competition of firsts.

• First to be held in the Middle East.
• First to use climate-controlled, open-air stadiums.
• First to be held out of season.
• First demountable stadium using recycled shipping containers in its construction.

After all, the World Cup and advances in television technology go hand-in-hand. Could it all be in the cloud by the next competition, in 2026?

Commentators in the US are suggesting the Superbowl (average audience in 2022: 112 million could be wholly cloud-hosted within 3 to 5 years.

It’s not completely unreasonable to believe the World Cup Final (average live audience for the 2018 final: 517 million, and live-to-air sports broadcasting more broadly, could do the same.

Media-specific cloud

It’s not news that cloud is the future. Episodic and cinematic rendering, some forms of post-production, and even distribution over the internet, have already made the move.

But whereas cloud-based content preparation for on-demand distribution is now commonplace, examples of large-scale deployment for media-specific use cases – such as playout encoding, multiplexing, and distribution to digital satellite, digital terrestrial, and digital cable – are very thin on the ground.

To date, broadcast distribution workflows in the cloud have been conducted in closed environments such as data centres, rather than the public cloud. But major sporting events need broadcast distribution that can scale to 100 million viewers worldwide – and that presents an economic, as much as a technical, challenge.

While lessons can be learned from accomplishments in other areas of broadcast media, live broadcast production and distribution have specific requirements both technical and commercial.

Broadly speaking there are three main pinch points to be considered: solution scaling; true virtualisation; and media SLAs. 

1.  Solution scaling

‘The cloud’ is often used as a catch-all term for various technologies, and open to interpretation. For most, the cloud is a useful shorthand to describe renting processing, networking, and storage capacity from a cloud service provider.

E-commerce and large-scale enterprise organisations have depended on and benefited from this model,and it’s why functionality like cinematic rendering and post-production are now almost wholly cloud-based. Accessing these services is one reason for live production or distribution workflows to move towards cloud.

But for true platform selling, a whole load of additional encoding, streaming, and packaging capabilities are needed. The main vendors are talking about these capabilities, but they remain difficult to scale. Digital terrestrial platform hosting, playout at scale, stream splicing, digital ad insertion, and certain aspects of digital rights management, still remain far removed from true cloud hosting.

In short, enterprise-level cloud provision has picked off low-hanging media fruit. The remaining, challenging areas of the market need a much more considered and industry-specific approach.

2.  True virtualisation

The same need for media-specific services is evident when it comes to virtualisation – or, put simply, distributed compute and storage. Popular with financial service and e-commerce businesses, virtualisation is also readily deployed in episodic and cinematic media sectors. But true virtualisation is not yet available for live, or distribution: there are too many media applications that purport to be virtualised but are basically the same old services with a different bow on top.

Because so many operations – ad insertion, localisation, overdubbing or audio description tracks – are tightly coupled with the broadcast, they really challenge cloud hosting. And although virtualisation attracts more than its fair share of cloud-adjacent buzzwords – such as Kubernetes, microservices, function-as-a-service – these tightly coupled media applications are not easily realised as microservices or containerised applications.

Software Defined Networking (SDN), device discovery, and Network Device Interface (NDI), among other technologies, will all need to be in the mix. As do automation and orchestration for deterministic media routing, to create a fully integrated management screen coordinating these disparate areas, protocols and SMPTE standards - rather than a hodgepodge of technologies.

3. Media SLAs

Service level agreements are the enablers of cloud provision. But once again, many of the SLAs associated with cloud hosting are designed for standard enterprise provision. They do not relate to media content, the availability of service, and the back-up for continual high-quality broadcast.

This has enabled generic vendors to carve out a lucrative position - taking on the more straightforward needs of media companies, overlooking more complex areas, and creating a labyrinthine commercial landscape in which media companies are left to calculate the commercial merits of various permutations of public, private, and on-prem deployment.

If large-scale broadcast production and distribution are to become truly cloud-capable, then more commercially viable and comprehensive SLAs will be needed.

No one should pretend cloud-hosting at scale is easy to deliver, or accept claims that it is.

The technological capability is there, but untried at scale. And the commercial model is more involved than the straightforward switch from capital to operational expense defining many large-scale cloud operations.

So, to return to the original question: will this be last cloudless World Cup Final? Maybe not. But if engineers and technologists have the ingenuity to create environmentally-sound, open-space air conditioning, they can certainly create a cloud platform for global-scale, broadcast distribution in the same timeframe.