The Challenge of Delivering Online Video: The Factors You Didn’t Know Affect Quality  


The world is witnessing a tremendous shift in the modalities through which media is consumed. The next time you are at a mall or restaurant, look at the kids with smartphones or tablets. Better yet, look at the adults on your train, bus or flight. You likely have at least a few devices in your home (forgetting for a moment smartphones and tablets) that are capable of accessing video content (Xbox, PlayStation, later model TV set, Blu-ray player, etc.). Media that entertains or keeps us informed is readily available from a variety of sources, in a variety of modes (VOD, near-live and live) and of varying popularity (one to many, one to a few, one to a very few) is now available anywhere and at any time.

Consumer demand is behind several trends today that stress a service provider’s ability to always deliver a high-quality experience, and they bear discussion:

The growth of the Long Tail:

In broadcast’s early years, choices were extremely limited. In the modern cable and satellite television era, the choice remains limited when compared to the outrageous variety of content available from sources such as YouTube, Hulu, Vimeo, and others.

This content is in demand, it is consumed and even generated in some cases by users, and it requires that service providers engineer specifically for the use case. A deep integration with an enterprise content delivery network (CDN), larger deployment of distributed storage, a “mid-tier” cache, and software engineered specifically for this use case (deduplication, intelligent chunking, etc.) are all requirements for successfully servicing this demand.

Live Event and Simulcast:

Eight years ago I worked with a New York-based terrestrial broadcaster on an early iPhone app for live access to the broadcaster’s programming. A small fortune was spent and leading edge techniques of the time were employed to deliver an experience that was limited in quality, costly to support and viewed by a very few.

Today, many of us have the expectation that we can find a live event online and view it in a reasonable quality. As cord-cutters, my family of four fired up the YouTube app via Apple TV to view this year’s US presidential debates, along with apparently ~1,999,999 other viewers (YouTube ended up supporting about 2 million concurrent live streams during the debates which were the most heavily viewed in US history).

When we catch up on the evening news, we select one of the network news apps available on the same platform to view simulcast content (favorites include CNNgo and CBS).

These use cases dictate another group of capabilities to succeed. Number one would be larger capacity commitments directly and indirectly with “eyeball” networks to support large “flash” crowds) to go along with the deployment of multiple CDNs working in concert with traffic-shaping technologies.

Everything Mobile:

While internet video has grown dramatically over the past several years, growth in mobile has been the leader. According to a 2016 report from Ooyala, 48% of all video plays in Q1 2016 were on mobile devices like tablets and smartphones, with tablet and smartphone video consumption growing 35% in the past year and 170% since 2013. These devices will inevitably connect via marginal or overtaxed resources such as public Wi-Fi, 3G, 4G, etc.

So we’ve looked at challenges specific to supporting long tail content, along with live and simulcast. The unstated assumption is that video on demand (VOD) of a “reasonable” popularity has become a well understood and “engineered for” task, yet mobile consumption adds a wrinkle to every use case.

Content types (long tail, live, VOD) can be accepted or not by a service provider as a business choice. It is clear in advance if we have the technology and resources necessary to support the content type. The consumption modality is entirely in the hands of the consumer.

Tablets and phones are free-floating devices that connect via Wi-Fi, 3G, 4G, etc. The only consistency is that none of these connections are consistent. With these consumers in mind, there are yet again specific steps that can be taken to provide.

Peering and transit specific to mobile carriers is a starting point, but an active role in preparing and transforming content adaptively (packaging, adaptive bitrates, and renderings appropriate to screen sizes are all a part of the mobile puzzle. Applications and “players” that employ alternative protocols such as UDP can provide an additional improvement in experience.

One thing is clear: As we increasingly replace “broadcast” with IP as a transmission vehicle for media, there is plenty of work still left to be done. The broadcast is a wonderfully efficient medium for the bulk distribution of content, but changing tastes in media entertainment (short form, USG, watch anywhere) coupled with the value of a two-way “conversation” with the consumer make a compelling case for continued growth in IP.

In order to fully realize the benefits offered in a shift to IP delivery, while containing costs, content publishers and media vendors must continue to engage in an ongoing dialogue with technologists in the network technologies space to grow in a scalable and sustainable direction.

John Cahill,
Vice President of Global Sales,
Highwinds CDN.


Please enter your comment!
Please enter your name here