SF VideoTech Talk: new streaming records

This Thursday, I gave a presentation at the SF Video Meetup on Peer5’s experience with streaming the 2018 Fifa World Cup. The video is going to be published in a few days, but here's some context:

The World Cup has been amazing so far, with tons of drama, surprises and excitement. For us it's doubly exciting because we're seeing a lot of traffic and gaining new customers. Because we perform better as traffic increases, we provide a huge value to our customers during events like this - in some cases we’ve been able to offload 90%(!) of a full World Cup broadcast, providing a 10x increase in scalability over what previously existed!

Is streaming mainstream yet?

The World Cup is also a good event to evaluate how the streaming industry is doing at a macro level. There’s a bigger shift that’s been happening for a few years now - the broadcast world transitioning / converging into the digital world. It’s not always clear where we are in this transition. The World Cup is a fantastic opportunity to look at the numbers and analyze. To understand where we are in this massive transition I compiled some data from various sources. To me, it boils down to two important questions:

  • Are we breaking streaming records?
    The World Cup is traditionally the biggest TV event, so if streaming is a real option for viewers, it should also crush all previous streaming records. I predicted that it would, some people said otherwise and the truth is that we don’t really know yet. But here is my analysis, based on the Conviva reported numbers:


  • How does streaming quality compare with broadcast TV quality?
    Companies have invested millions of dollars in their streaming infrastructures the last few years to prepare for the World Cup. Now is a good time to see if that investment has paid off. The short answer is no, as people are still complaining about buffering and other quality problems.
How does Peer5 measure quality?

quality metrics

We track a variety of quality metrics but let’s focus on just 2 for this discussion. The first 2 graphs show buffering % - i.e., time spent buffering divided by the sum of (1) time spent buffering and (2) time spent watching video.

The 1st graph shows the buffering chart for a broadcaster that operates its own homegrown CDN. With these clients we typically see a buffering % greater than 1% with spikes when the servers are overwhelmed.

The 2nd graph shows the buffering chart for a broadcaster that uses a single commercial CDN for video delivery. Assuming the CDN has good POP coverage in the broadcaster’s primary geography, we typically see less than 1% buffering for these types of clients with the occasional spike when something bad happens.

The last chart (% of viewers who experience a rendition switchdown) provides a different view into quality and is useful for all types of architectures, but especially so for broadcasters who employ multi-CDN delivery. For such clients, we typically see very little buffering. But, does this mean that viewers are getting high quality streams? Not necessarily, as this chart indicates. If a high % of viewers are being transitioned to a lower quality level, this indicates that at least one CDN in the portfolio is struggling, even if buffering is minimized.

How is Peer5 is doing so far?


Here are the slides:

Peer5 help video broadcasters such as FuboTV, Turner and Grupo Clarin to ensure perfect video quality. If you’d like to see Peer5’s P2P video delivery options for yourself, feel free to take our free trial!