Low-Latency Sports Streaming at Scale

Learn more about low-latency streaming at Streaming Media's next event.

Watch the complete presentation from Streaming Media West, ESS202. Moneyball: Monetizing Your Assets, in the Streaming Media Conference Video Portal.

Read the complete transcript of this clip:

Steve Heffernan: We can, for the most part today, reach low latencies, even at the scale of big sporting events. The problem is, it's very expensive. So like we look at a game like HQ Trivia, right. When they were first spinning up, getting popular, they were still streaming RTMP to every single viewer of that app. And we were doing the math on that, like assuming they're were using a Flash Media Server Wowza. They were spending $20,000 just on encoding for a 15-minute game. All in, you're looking at $50-100,000 for a 15-minute segment of game.

So, coming back to betting, we were saying that there was not a lot of profit in betting. Like that's a lot of that profit going out the door, for such a short period of video, right? So that's where this push is coming from to use HTTP-based technologies as opposed to RTMP or WebRTC, two protocols that are very expensive to scale up. HTTP is this protocol that CDNs have been optimizing for 20 years now, is using commodity hardware on the edge, it's 10 times more expensive to stream on one of these other protocols than it is scaling up HTTP.

I'm sure you're familiar with HLS and DASH and those kinda options that you have for doing this adaptive streaming. Well those two platforms, I guess you might say, are both kind of exploring how they can get their mechanisms down to lower latency. And there's been one in process for over a year now on the DASH side bleeding into the HLS space.

And then, as Brian mentioned, a couple of months ago, Apple came out with their own version, which is actually completely different from what the community was doing. So maybe quickly to speak to those options, the first one, the DASH option is basically using a technology that allows the player to request the next part of video before it is even created. So you're saying, "Give me the next segment of video," and it's kind of like hanging, and then like as the video's coming in, the bytes are coming in, the player's able to shove it into the player as the bytes are streaming.

The Apple approach is using a newer technology with HTTP--HTTP/2--has a technology called Push in it, and this is mainly used for... So, thinking of an HTML page, and that HTML page has CSS files and Jobster files attached to it. What this mechanism does is, it allows the server to say, "Oh, you're requesting this HTML page, I'm also gonna give you these other things as well."

And so, what we're using this in video, what Apple is proposing to use in video, is to say like, "Okay, well you're asking for this manifest, but I'm also going to give you the next segment of video in that." So it's kind of like a different approach to do the same thing, like getting the files from the server to the client without more communication happening and these pauses in between.