Online TV Case

How to transform the quantity into quality and decrease buffering 50%

58% of respondents of the most recent Bitmovin Video Developer Report said that good performance means low buffering rate. We all know how annoying it is to wait until the rolling cycle in the video player finally disappears. But why does buffering ever happen? And is there a way to reduce it without changing anything on the backend of your delivery infrastructure?

With this use case, let’s discover how it’s possible to influence buffering on the last mile of delivery. We’ll share our customer’s impressive metrics with you. But first, a piece of basics.

Where does the buffering come from?

The main cause for buffering — a slowdown in the download process. When the next chunk isn’t yet available for replay, there’s no choice other than to lower bitrate or (when this doesn’t work out) to freeze the playback until the buffer fills again. So it’s a bandwidth issue in the first place, but is this really the norm?

For decades it was a “game” with just two sides: the downloader (the viewer) and the uploader (the CDN). The effective data transmission rate couldn’t be higher than the lowest bandwidth available in one of these two points. And it seems like no matter where exactly things go wrong, on a CDN side or on a viewer’s side, if we can’t somehow eliminate the bottleneck, the buffering remains.

Distributed CDN means multiple traffic sources

Everything changes when a video player gets multiple connections to a distributed cloud of devices and maintains them during the entire session. Of course, some devices could go offline (just as some CDN edges could become unreachable), but if a cloud is big enough, every viewer has many active nodes to download traffic from.

On the X — a number of channels used in a single viewing session, on the Y — a number of viewers that have that particular amount of active channels during the session.

This leads to a download speed increase

In usual circumstances, it would inevitably lead to bitrate downscale, and then to buffering at some point. But having many active channels actually fixes that problem! Statistically, we see that viewers’ devices are able to download the video several times faster from the other devices in a distributed cloud than from the CDN!

The chart shows a group of viewers that experience buffering. And how faster an active channel could be compared to CDN download speed. On the X — 1 means equal CDN and P2P bandwidth, all other numbers mean how many times the download speed from P2P exceeds CDN speed in the same viewing session. Logarithmic scale.

This results in buffering rate decrease

Do you remember — faster the download, lower the chances of buffering? It’s never zero simply because some devices are in such extreme conditions that even a few traffic sources are unable to deliver video immediately. However, statistically, we’ve got that P2P CDN has a positive effect on buffering.

These curves are for 2 groups of viewers during the same 24-hours on the same website. The green line is the buffering rate for a group connected to CDN only. The red line is for a group connected both to CDN and distributed delivery network.

If a viewer has a set of active channels to download from, the chances of buffering fall 50%!

Being on the last mile of the supply chain, we believe the critical thing for keeping QoE at the highest level is the ability to detect video download problems and the ability to change the way a particular device is downloading the file in due time. When every single video player has multiple live connections with multiple distributed sources, there is always a lightning-fast way to switch from a bad source to a good one and avoid the download slowdown which usually causes buffering.

Andrei Klimenko, CEO Teleport Media

Keen to learn more in detail? We’ll share more insights with you!
Contact us today to start your free trial with unlimited traffic.