Blog
Video Streaming: A detailed look at traffic spikes! Part 2 of 3
Consumers’ insatiable appetite for videos over the internet is undeniable. Live prime-time events are increasing in frequency and in bandwidth demand. But is today’s network technology robust enough to handle this growing demand for online videos?
This is part two of our blog series “Video Streaming”, in which we explore the OTT video growth explosion, and look at how network operators can arm themselves to deliver high quality videos streams, reduce churn, and attract new customers. The first blog post can be found here.
Part 2: OTT Video Consumption Peaks
In the first part of our blog, we talked about the astonishing increase in OTT video consumption over the last few years, and how it will continue to grow in the future. In this second part we examine the peak traffic phenomenon. Traffic spikes, caused by millions of users watching the same content at the same time, during popular TV shows, premieres, breaking news or sports, are on the rise, both in frequency and bandwidth. Breaking news makes up 56% of most-watched live content and concert/ festivals are in second place with 43% (stats from Livestream and New York Magazine Survey).
Let’s look at some recent peak traffic examples. During 2019 Rugby World Cup, live streaming broke a new record in Africa during the final game, reaching a peak of 500 Gbps for the first time ever on the continent. During the 2018 US mid-term election there was a spike of 217% viewership. During the 2016 European Soccer Cup, UEFA enjoyed a one-day traffic increase of 128% on June 14th, reaching more than 5 million viewers.
These spikes were hard to predict. By the end of the Round of 16, after only 20 days of play, the EURO 2016 site had attracted around 89 million worldwide visits. This figure is higher than all previous visits combined since the beginning of 2016. https://www.similarweb.com/blog/euro-2016.
The FIFA World cup 2018 lifted global traffic up 29% according to Conviva’s State of the Streaming TV Industry 2018 report. Meanwhile, a large mobile operator in the UK calculated that during major sport events, mobile traffic could spike to more than 4 times their regular YouTube traffic. Telstra in Australia had seen large increases year-on-year in their live sport mobile traffic, e.g. 2.5 times increase from 2016 to 2017. At the time, live sporting events were transmitted using traditional OTT / Adaptive Bitrate streaming, i.e. in Unicast. This traffic increase was primarily driven by traffic spikes during particular matches.
Super Bowl LI, the American Football annual championship, matched the New England Patriots and the Atlanta Falcons. The carrier of Super Bowl LI, Fox, used a multi-CDN strategy to deliver the live streams. Yet the quality of the live stream was not perfect, and for a brief time many users were locked out of the stream as a result of technical difficulties. The feed didn’t meet linear TV’s so-called “five nines” reliability standards (no hitch 99.999% of the time).
Linear TV in traditional CDN deployments uses a single stream per user from the edge cache, even though many users are consuming the same content. This creates a scalability challenge for network operators.
Adding to increasing peak traffic, telecom operators have to deal with users who are more and more demanding for a great Quality of Experience (QoE). When watching a livestream broadcast, video quality is the most important factor for 67% of viewers (Livestream and New York Magazine Survey). Yet, video streaming still suffers from bad quality.
A good example of this occurred during the 2019 Rugby World Cup. New Zealand Telecom Operator Spark’s transmission of the quarter final, the All Blacks versus Ireland, was plagued with freezes and rebufferings. Furious subscribers lashed out at Spark Sports; Sparks’ Facebook page was flooded with complaints.
Viewers lose patience when faced with streaming TV delays as shown by the global 7% year-over-year increase in the rate of abandonment – which translates into 14.6% of viewers leaving before the video starts (2018 Conviva State of the Annual Streaming TV Industry). According to Convivia’s report, customer abandoning video viewing due to a bad QoE increased 16.7% in 2018 for live content in the USA.
Latency in OTT is a growing issue as well. Standard linear broadcast latency varies between 3.5 and 12 seconds, with satellites capable of being marginally faster than cable. OTT latency, however, is currently hovering around the 25-40 second mark (Viaccess-orca blog).
Video QoE issues cost operators both revenue and viewers. Recent research shows that there is a direct relationship between re-buffering and abandonment. Each instance of re-buffering results in a 1% abandonment rate. Improving video QoE also reduces churn, by 90% in one SVOD provider’s case.
With consumer demand on the rise, unpredictable peaks, and viewers’ high video quality expectations, what are network operators to do? We will explore an efficient solution, already implemented by a few operators worldwide, in our next blog.