How Broadcast and Social Complicate Streaming Latency
See more videos like this on StreamingMedia.com.
Learn more about streaming latency at Streaming Media East Connect 2021.
Read the complete transcript of this clip:
Eric Schumacher-Rasmussen: Tell me about latency and why it's so important for live streaming. Why it's not a one-size-fits-all proposition. I mean, it would be nice, wouldn't it, if we could just keep it simple and provide every end user in every situation with the same latency, but that doesn't make sense for a number of reasons, right?
Neil Glazebrook: I think you hit on it in your introduction about economics. But let's just back up a bit. And I'm certainly not going to take credit for this notion of the latency spectrum. I think our friends at Wowza and Mux have done a really good job of describing the overall latency choices that we, as technology companies, deliver these experiences to end customers. And I look at this latency spectrum--at one extreme, we have the standard chunk-delivery formats that we've known and loved, probably for a decade now, HLS and DASH, and we've come to expect what we would call hand-waving latency. That's the elapsed time between somebody standing behind the camera or waving a hand and somebody at the other end, that consumption device, the video playback device, being able to see that same action.
So we've come to accept latencies of around 45 seconds to a minute on those traditional streaming formats. And it works great. We've spoken about streaming events to tens of thousands, millions of audiences. I know that at the other extreme over recent years, we've seen near-realtime streaming latencies of less than a second, with technologies like WebRTC. And then we've seen kind of in the middle of reference point of broadcast, which would be digital DVB or cable. That hand-waving latency is anywhere between 5 and 8 seconds. So I think what we're seeing, in the latter half of 2020, and what we will see through 2021 is adaptations and enhancements of our streaming technologies to align with broadcast so that we can synchronize the experience of second-screen, for example, for live events. We allow interactivity with social media and so forth. So it would be great to give everybody a one-second or less-than-one-second latency experience.
That said, there are probably consequences to that. But one of the big ones, of course, is the economics. So I think we're looking at trying to optimize around that midpoint. Todd, did you want to add anything there?
Todd Erdley: Yeah, I have to agree. Let's blame it on broadcast. Let's blame it on social. And I think that Neil captured it perfectly because broadcast has becomes that gold standard, that five seconds. And then, things like this: I'm watching TV and then I'm going to be sending text messages to people about what I'm watching, and then they're going to consume it on a second-screen device and Neil's watching it as broadcast. And all of a sudden we're out of sync, and those are just experiences that have got to go away. And so we need to really look at live streaming no longer as a second-class citizen, but as a first-class citizen. And I think that the industry, be it in HTTP, be it in SRT, be it in WebRTC, be it an even in slice-based encoding, I think that we're really progressing things really well to drive that performance, to be a first-class experience.
Videon's Todd Erdley, AWS's Neil Glazebrook, and Streaming Media's Eric Schumacher-Rasmussen discuss the potential role of RTMP in low-latency delivery in streaming's post-Flash present and future.
Companies and Suppliers Mentioned