When networks hit the wall


Whether you are streaming the latest boxing match or watching the most recent Game of Thrones episode, during its first run, the network is going to play a major role in determining your quality of experience like never before. There is unprecedented demand for content, and with the proliferation of smart devices capable of displaying video, every pair of eyeballs on the internet is a potential consumer.

The widespread availability of video combined with streaming technology means our viewing expectations are now much more demanding on the infrastructure. We expect content to be there, wherever and whenever we want it. We want to watch it, pause it, play it in slow motion, watch it again, analyze it, save it for later, share it with friends.  We want to watch it at home, on the train or at work – because there’s always the underlying risk of spoilers, or missing out on the next-day office discussion should we be unable to view it.

There have been some very visible failures when streaming high demand content. Just last month Game of Thrones fans expressed their disappointment about the season opener. The viewing experience just wasn’t the way many anticipated. Due to incredible interest in the 7th season premiere, many viewers in the Americas decided to stream the season opener since the stream is made available simultaneously, regardless of time zone. Some ended up suffering through a failed website and were subjected to a streaming experience full of delays and buffering. An experience that just didn’t meet expectations for a show with such high production values. They took to social media outlets to share their disappointment and thoughts on the matter. Even more recently, the poor streaming experience that some viewers had during the pay-per-view Mayweather-McGregor fight has resulted in legal action being taken.

But could these quality-of-experience failures have been avoided? You may look at other “over-the-top” (OTT) services and wonder why these problems haven’t been wrung out of the infrastructure by now.  There are clear challenges when live-streaming an event. Live coverage requires constant, uncongested connectivity and widely available capacity. Feeds can be made multiple redundant and traffic engineering tools used to provide the highest probability of success. Even then, the demand for any given event is notoriously hard to predict, and if a live event does unexpectedly “go viral” congestion, buffering delays and outright overload and system failures can occur. This is particularly true in static legacy networks where the data pipes are only so big and there is significant oversubscription at the network edges. The underlying technology just wasn’t designed for this level of streaming.

Leave a Reply

Your email address will not be published. Required fields are marked *