How Quortex Cloud Streaming is green
Getting sustainable streams is a long and winding road. We don’t pretend that we will change that in a snap, nor do we pretend that streaming is perfect by all means. But sustainability is a mix of small things, break-through innovation and big ideas with no compromise on OTT experience. Let’s crack it!
Quortex, founding member of Greening of Streaming
We joined the Greening of Streaming alliance as a founding member. This alliance is made of industry experts of all horizons. There are many different goals to that alliance, but one of them is to draw the big picture of the streaming industry. Is streaming worst or better than broadcasting? What is the relative impact of the CDN, the player, the TV set, the transcoding or the ingest? What kind of energy is used by the cloud providers? Drawing this picture is of paramount importance as we need to focus our efforts on what makes a difference.
Just-In-Time, an innovative approach that reduces the streaming impact
One of the only principles that everyone agrees on is that the greenest energy is the one we don’t consume. It should be a clear goal to lower the waste to its bare minimum, without impacting the quality of service for the end user.
The streaming industry architecture was mainly inherited from traditional broadcast models, where a “push” approach makes sense: the content is pushed to the receiver in a “one to several” model. While streaming is clearly a pull model (you make a connection to a server to get your files), this legacy model is still widely used, and all the different flavors of an OTT stream (profiles x packaging x encryptions) is pushed to the CDN, regardless of the actual demand. It means that the user specific needs (in terms of bitrate or resolution) are ignored, but it also means that a huge amount of resources (and kW.h) are used while no-one may actually be watching the content! In other words, you heat your house and leave all the lights on even though no-one is at home.
The approach we are having is the exact opposite: we provision only the exact amount of CPU/Network/Storage resources that is required for the service at any time of the day. If only a single profile is pulled, we won’t process any other ones. If a channel is simply not watched, we will get rid of any associated ressources; and if the demand is skyrocketing, we will automatically adjust. In other words, we automatically turn the heat and the lights off when no-one is in the room! This translates into energy savings (and also into cost savings by the way!).
Using wasted resources
It’s obvious that a clever way to save energy is to use unused resources. That’s one of the challenges of electricity distribution: most of the time, the produced energy is wasted as it can’t be stored when the production exceeds the demand. The same applies to cloud computing: providers need to build huge data centers to accommodate for peaks, which means that servers are not being used (but still consume electricity) most of the time. For that reason, they created the “Spot” instance concept: these are resources that may or may not be available (depending on the demand) and that are priced at a fraction of the nominal price to encourage people using them (hence maximising the resource usage and making the best use of the spent kW.h)
At Quortex, we developed a very specific live streaming service that can leverage these spot instances. We will always try to use them in priority and 90% of our workflow runs on Spot. It translates into cost savings for our customers, but it also translates into a better usage of electricity. Just think of it as if we had the ability to use the heat from our neighbor’s heaters that left them on while they were not in their house!
