Why Large-Scale Streams Fail

Learn more about streaming at scale at 流媒体东部 2022.

Read the complete transcript of this clip:

杰森·Thibeault: 通常, when we talk about scale and video, we're just talking about, can we deliver bits to an increasingly large number of concurrent users? 好吧. 那太好了. But there's so much more involved with live streaming an event than just delivery of bits. So, things like authentication, portals, things like 支付网关. What are some of the other things that a streaming operator really needs to keep in mind to scale, as they're thinking about expanding the number of concurrent viewers they have? What are all the other pieces besides just delivery?

比尔Wishon: I think that when we talk about scale, we ought to think about two parts of the scale. One is just the overall concurrency you expect, but the other is how fast you expect your audience to come online. Because we've seen failures in both areas before, if you go back and look at examples of things that have failed, you can see examples of things that have failed because the rate of transactions was too high. 例如, where you have a payment gateway that's in the process of people getting on board to the stream, they have to pay for stream. And it gets inundated with a surge of people wanting to join the event that takes down the payment gateway, takes down the entire event, 本质上, even if the streaming infrastructure was going to be perfectly fine in terms of when the user finally joined, you had an issue at the in the workflow of the user getting online that wasn't properly examined and scaled and perhaps not even a backup plan even planned, they had a single point of failure.

And so anything that's in your user flow from when they get aware of your event to when they come online to whatever link they click to get to the first place where they're gonna then either subscribe or put in their email or whatever it is that their flow is. You gotta look at all of the pieces that are being touched, whether it's authentication systems, 支付网关, other systems like ad insertion systems along the way, and whether they scale to rate of transaction and at stable state the overall audience size.

Bulldog DM's John Petrocelli Talks Post-Pandemic Streaming at Scale

流媒体's Tim Siglin and Bulldog DM's John Petrocelli discuss how tentpole live streaming events have changed over the last three years for the streaming pros who deliver them, the audiences that engage with them, and the brands that invest in them in this interview from 流媒体东部 2022.

2022年6月6日