Netflix Streaming vs. Reality: Long Aggregations are Misleading

Woman watching online streaming service

by | May 25, 2017 | Networking

The Reality of Netflix Streaming

Due to a nasty flu that hit my family last weekend, our household had an overdose of Netflix streaming. Looking at the Preseem charts from those days, I was reminded of the value of metrics sampled at high frequency.

To illustrate this, let’s first look at the average bit rate during one of the flu-induced binge sessions:

Screenshot of Netflix Downstream Bitrate

Downstream data using Preseem taken from Dan’s home.

Due to the time range selected, the above chart uses a 218 second (3.6 minute) interval. This fits the bandwidth profile of Netflix streaming, showing a relatively constant bit rate.

Preseem is based on 10s samples. When we do aggregations for longer reporting periods, we calculate the aggregate value but also keep track of the peak value in the aggregated time intervals. Let’s look at the same chart with the peak values overlaid on top of the average values:

Screenshot of Netflix Downstream Bitrate

Downstream data using Preseem, taken from Dan’s home.

Are you surprised? My home has a 25 Mb/s connection and, over short intervals, Netflix consumes most of the available bandwidth. This is the case despite the longer-term average video stream rate being only 4 Mb/s.

What does this mean for your subscriber’s quality of experience? Since the Netflix streaming was only 4 Mb/s on a 25 Mb/s connection, you might conclude that there’s 21 Mb/s unused. Which means the customer won’t have any quality problems, right? Unfortunately, that’s not true. Each of those spikes could have been enough to cause lost packets in a VoIP call or add lag to a gaming session.

If you’d like to learn more about how Preseem can help improve QoE for your subscribers, click here to book a demo.

Subscribe to the Preseem Blog Newsletter

Stay in-the-know and get fresh content delivered to your inbox once a month.