How Disney and EvocaTV Approach Streaming QoS and QoE

How do streaming platforms such as Disney and EvocaTV handle the variables of Quality of Service (QoS) and Quality of Experience (QoE)? Nadine Krefetz, Consultant, Reality Software, Contributing Editor, Streaming Media, asks Pankaj Chaudhari, Architect, Video Delivery, Disney Streaming, and Imran Maskatia, VP of Product Development, Evoca TV, to discuss the ways their organizations work to improve both QoS and QoE and what challenges they face in the process.

“Our focus is always from the client-side point of view,” Chaudhari says. “The QE metrics, your typical metrics on play, start, play failure, rebuffing, and such. But they’re also a function of a root cause that is happening or that has happened in the past upstream of where the player sits. And so it is absolutely important to actually figure out where these problems happen, so you can react to them much sooner. And these systems are generally your CDM delivery side of things, your channel signal acquisition, but then also your services layer. So the player first starts talking to your services layer running in the cloud, let’s say, [for] those systems to be up there running without much latency, and they should have availability. And once that handshake completes with the services layer to get access to the playback URL, then the player starts talking to the CDN and should start pulling the content.”

Chaudhari further elaborates that the approach is different for live streaming and requires other solutions. “In a live case, the content is never sitting with the CDN except for the content age is out really fast because everybody’s generally at the live edge,” he says. “So the CDNs invariably have to keep going upstream to the shield, to the origin, but the origin also never has the content. So it has to be coming in from the signal. And so having visibility into these different systems and having a DevOps model and owning the responsibility of different systems in your pipeline and making sure that availability and latency is under check, and if they pass a specific threshold, you should signal on those.”

Chaudhari goes on to say that all these variables can impact performance on the player side, and they require close vigilance in order to minimize problems. “When those thresholds add up, it’ll have a negative impact on the player side,” he says. “So it kind of becomes important to monitor when a specific system starts to flag…and then alert on that. And then based on that, you can actually have your mitigation strategies as well, too.”

“So realistically all of you guys have something, I’m going to assume, for QoE, QoS,” Krefetz says. “Are there issues with how it works for you or is it working fine?”

“It could always work better,” Maskatia says with a laugh. “I don’t think you’re ever done with QoE. You’re never just, ‘oh, we’re done!’’

“It’s not good enough?” Krefetz says. “You’re a perfectionist?”

“It’s about getting visibility to as much of the telemetry as you can and then understanding it,” Maskatia says. “Because at some point when you’ve got so much data, it’s about really trying to put it in the format that you can visualize and really pinpoint easily where we’re having a problem in, you know, this particular CDN node in Baton Rouge, Louisiana, or something like that, right? So there are issues like that. And then the entire service layer needs to be monitored, you know, to [Chaudhari’s] point, this is a DevOps model…the bane of existence to video services is not always the video. Usually, it’s nine times out of ten, there’s an authentication issue, right? Or there’s some other issue that’s not even in the CDN delivery. And so you’ve got to really monitor all of the services involved in delivering video as opposed to just what delivers the video bits.”

Learn more about QoS and QoE at Streaming Media West 2022.

Watch full-session videos from Streaming Media Connect 2022.