First, the Oscars are a streaming disaster, than HBOGO matches the streaming blunder with True Detective. Some have blamed the outage on traffic overflow. While many in the industry might believe this, I certainly don’t. The Olympics weren’t a disaster, and neither were other big live events.
There’s a lot of moving pieces in live event delivery, and if one piece fails, than everything fails. In my experience, 99% of live event delivery failure is due to the human factor, not the infrastructure. I can almost guarantee that if the Oscars and True Detective were completely outsourced to Akamai or Level 3, the issues would have been avoided.
For the Oscars, if Uplynk was used as the intermediary platform, than the blame is squarely on them. Uplynk provides an end-to-end video delivery service, from ingest to consumption, and everything in between. Uplynk is able to ingest one high quality video feed at the customer location, then take care of the rest, pushing the stream to the cloud, where it is transcoded to different renditions, and finally delivered to any device anywhere in the world, via Amazon CloudFront.
There is only one big issue. The Uplynk platform uses Amazon CloudFront for video delivery. Not even Netflix uses Amazon CloudFront for video delivery, but instead uses Akamai, Limelight and Level 3. Now that Uplynk is part of Verizon, the bad press is on Verizon. Everyone is the industry understands that CloudFront is not good for streaming VOD or live events, especially on a large scale.
Video streaming requires dedicated high powered infrastructure, not shared infrastructure. Akamai and Level 3 are specialized for this type of delivery. These two CDNs have been doing large scale live events for many years. Amazon is new to live streaming, and its going to takes years for them to figure it out.
What is Verizon to do with Uplynk
There are three things that Verizon needs to do, to get this right. First, Verizon needs to drop Amazon, and build a private cloud, where Uplynk is running on its own dedicated server infrastructure. There is nothing better for high performance streaming than running it on high performance dedicated servers, that have been optimized for live events.
Second, Verizon needs to switch over from CloudFront to EdgeCast. EdgeCast now has the infrastructure in place to do large scale live events. Third, Verizon needs to use hardware based encoders like the Ateme or Media Excel, for onsite live video ingest. These encoding workhorses never break down. They can ingest one high quality video feed, and spit out as many renditions as necessary, without breaking a sweat.
These two live streaming event failures are giving the whole CDN industry a bad wrap, especially when Dan Rayburn is saying that Internet (specifically CDN infrastructure) is not yet able to compete with the linear broadcast infrastructure. To that, I say please, live streaming is ready for prime time. Don’t blame the CDN infrastructure, when the failure was do to human error, and the wrong pieces being used.