Why Streaming is Still Struggling With Real-Time Delivery
Why are we still talking about the challenges of low-latency streaming? Put another way, with all the hype about real-time streaming and its evident advantages for high-interactivity applications ripe for monetization like iGaming and sports betting, why haven't we made more progress toward realizing real-time and its potential?
Allan McLennan, President, 2G Digital Post Optimization, discusses this topic with Ryan Jespersen, Director of Product Strategy, Dolby.io, Corey Smith, Sr. Director, Advanced Production Technology, CBS Sports Digital, Paramount, and Evan Statton, Sr. Principal Solutions Architect, M&E, AWS, in this clip from Streaming Media Connect 2023.
McLennan says that platforms such as Twitch have been phenomenal with efficient real-time streaming. However, he mentions that delays of even one second are unacceptable in areas like live sports, esports, and sports betting. He asks Jespersen, “Why haven’t we moved further?”
Jespersen says that, while this may be considered a bit controversial, everything in streaming is a hack. “Even my own startup, we hacked WebRTC to make it a service for linear real-time streaming. WebRTC was never built for that. It was built as a real-time communication platform to build Google Meet and [what used to be] Google Hangouts.” He says that since it was available on every browser and many devices, “You can do it from contribution and distribution. And we created a service, and we were the first to do that. And that's true of pretty much anything that's out there.”
Jespersen says that once Flash RTMP disappeared, all the CDNs “got rid of their stateful architecture. They all used to run origin edge CDNs that ran RTMP servers, and they would do an origin ingest, they would pass it to an edge, and the edge would fully connect to the Flash clients, that would be the players. And that was an RTMP origin edge. And that's what we are doing with WebRTC, and it's very hard to do. And more importantly, it's very expensive because you have to spin up actual servers to create those stateful connections that map it. The benefit of all of is that you get the [advantage] of two-way data channels. You get the benefit of there being interactivity in the streams that you're doing. You get a benefit to a whole range of other things, both in the days of RTMP and in the days of WebRTC now, where you can now create watch party type use cases or breakout rooms or other forms of interactivity that we haven't even come up with yet that I think are going to be very interesting as we try to figure out new ways to generate revenue and do direct to consumer applications.”
This is why the HTTP streaming industry has essentially taken the most inexpensive route for optimizing live streams, Jespersen says. “We have this amazing HTTP architecture that scales all of our other websites and everything else, and HLS was a brilliant protocol for making the iPhone ubiquitous and for making streaming ubiquitous. So I'm not piling on HLS because it was such an amazing hack of how can we use HTTP and HTTP CDNs to scale streaming and make it accessible to massive global audiences. And to this day, there's nothing in the WebRTC space or in the RTMP space or in any space that can get to the levels of scalability that HLS enables us to. But if you have a use case that requires anything under five or 10 seconds, it's really hard to get to a massive scale on HTTP and have extremely low latency. Because of the way that HTTP CDNs are distributed, you've got to cache out through all these different layers of edge architectures to get packets all the way down. It's really hard. Despite the encoding technologies being there with Common Media Application Format (CMAF) and Low Latency DASH (LL-DASH), it's really hard to proliferate those packets across these architectures to deliver those in time that don't have this inherent latency. That's kind of the simple answer to why HTTP is struggling to deliver on the promise of real time.”
Corey Smith says that the industry needs to come up with more of a “Swiss Army Knife-type of solution.” “I think we've kind of lost our way a little bit in terms of just piling on HLS because ‘Hey, it just works, right?” he says. “But then we also talk to our CDN providers, [and] they're not willing to do anything because everybody wants to be on HLS, so they have no incentive to change up their architecture to move anywhere in the future.” He wonders why, since all of these technologies are essentially computing at the edge, no one is building products that would be “de facto on the edge.” Ultimately, he emphasizes that the streaming industry has “the power to influence how stuff is being delivered to the clients in a more cost-effective and more of a higher efficient way than just saying, ‘Oh, well we'll just move WebRTC because it's fast and has low latency. I mean, to switch up infrastructure like that is not something that can be done overnight. But at the same time, we can't just keep hopping from different format to format.”
Statton says that he likes Smith’s idea of developing generic computing at edge. “There's a bunch of computers out there doing something, and right now, they're all pretty differentiated in what they do,” he says. “I think the lower the latency you can get for live events, the more monetization opportunities you have, all the way up to real-time betting. There's some stuff you can do under a second. There's some stuff you can do if it's within a few seconds. There's some stuff you can do if it's within a minute. And then the value of the content sort of deteriorates. I'm specifically [about] live sports and news. The further you get from the live head, the less value that content has almost immediately. It's like when you buy a new car, then you drive it off the lot. So you need to have all this technology and all these options to give as much experience to different users who want different experiences as you can so that the rights holders can have more monetization channels.”
Watch full sessions from Streaming Media Connect November 2023. We'll be back in person for Streaming Media NYC on May 20-22, 2024. More details here.
Related Articles
With large sports-streaming operators, WebRTC provides a real opportunity for ultra-low-latency streaming. But those same operators, which spend billions on licensing rights, can't afford to just swap the ability to stream content in real time for basic OTT functionality like SSAI and DRM.
19 Jan 2024
Parks Associates VP, Research Jennifer Kent discusses the forces driving ultra-low latency streaming including the ascendant sports betting industry in this interview with Streaming Media Editor-in-Chief Steve Nathans-Kelly at Streaming Media West 2022.
02 Jan 2023
Dolby.io Director Product Strategy Ryan Jespersen discusses Dolby's WebRTC-based scalable real-time video delivery technology in this interview with Streaming Media's Marc Franklin on the show floor at Streaming Media West.
08 Dec 2022
Red5 Pro CEO Chris Allen discusses the growing market for ultra-low latency, interactive streaming in applications like auctions, betting, and video conferencing and a multi-cloud approach to delivering it in this interview with Streaming Media's Tim Siglin at Streaming Media East 2022.
14 Jun 2022
Companies and Suppliers Mentioned