Server-Side Selection is a Game Changer for Video Streaming
The video streaming market is growing rapidly. A recent report from Allied Market Research found that the global video streaming market will reach $149.34 billion by 2026. While the rise in video streaming represents an opportunity for revenue growth, it's also a challenge, especially in terms of providing the best quality of experience (QoE) on all devices.
Today, adaptive bitrate (ABR) technology delivers several quality streams of the same content, leaving it up to the end-device to decide which version to use. As ABR traffic soars, driven by the increasing number of mobile devices and by the progressive adoption for linear channels on big screens in place of traditional broadcast, operators are losing control over quality and resource usage. This puts operators in a situation where they are no longer able to guarantee a minimum quality of experience for their TV service at all times, even for high-value content.
Limitations of ABR Streaming
To date, ABR has been working well because it's still in the early adoption phase. If ABR was applied to all video consumption, then that would put an unprecedented load on networks, which could easily exceed their current capacities. This is especially true for linear channels, which represent about two-thirds of video consumption and are still mostly received via dedicated broadcast distribution methods today. If linear channels were all distributed via ABR that would triple the current amount of ABR traffic. In reality, the volume of traffic would probably be even greater since network dimensioning is based on peaks rather than average usage.
The huge and unpredictable scaling efforts required to deliver popular and high-concurrency content is a major challenge for ABR streaming, and a rationalization of the delivery will become an absolute necessity in the future as video traffic increases.
ABR streaming standard implementation relies exclusively on the players to evaluate bandwidth and perform quality selection. The problem with this approach is that each device acts as if it were alone on the network, with no considerations of the other devices sharing the same resource and no considerations regarding the state of the network.
Many players typically use the highest quality stream available on the network, regardless of the screen resolution of its hosting device. This ends up wasting bandwidth on smaller displays such as smartphones where the perceived video quality would have been the same with a much lower bitrate.
Moreover, when devices are competing for common resources, there's no rules to determine how to split the resource between the different devices. This generally results in an unfair, unpredictable, and unstable distribution of resources. The fight for resources can often lead to unstable fluctuations for who is getting more bandwidth at a given time and significantly impact the end-user experience.
These issues are further amplified when trying to reduce the extra latency that is generally implied in ABR. ABR content typically appears on the display 20 to 30 seconds later than what people are used to with traditional broadcast. Solutions exist to get rid of this delay but they mainly rely on video segment chunking, and this operation makes the job of stream selection much more complex for players that only more obviously disclose the instability problems described previously.
For ABR to be able to provide a quality of service equivalent and even better than TS-based solutions, these issues need to be tacked and solved. And this goes through an increase control on ABR traffic.
How to Better Control ABR Traffic
Operators require more control over how their network is used by ABR devices for at least two key reasons. First, to ensure proper usage of the infrastructure they're investing in. Second, because it's the only way to make certain that high-value video streaming content is delivered with a minimum guaranteed quality to all users.
To ensure that bandwidth consumption always remains within a determined limit and manage how resources are shared between users, server-side segment selection streaming technology is a radically new and innovative approach where video stream selection is not solely left to clients individually but also controlled server-side by the CDN.
With server-side segment selection streaming, CDN servers are able to assess in real-time the available bitrate for every connection established with an end-device and to dynamically alter the resolution choice initially made by the player (downward as well as upward). This choice can be based on the measured bitrate above but also on external parameters, such as pre-defined rules or commands, in a way that doesn't affect the ongoing playback on the devices.
The selected streams sent from the server to the end devices remain fully compliant to streaming standards, including Apple HLS, MPEG-DASH, and Microsoft Smooth Streaming. They work seamlessly on any player without requiring any type of specific adaptation, which greatly simplifies the implementation of this technology.
Benefits of Server-Side Selection
Server-side segment selection allows operators to decide which quality to deliver to each connection based on network-aware criteria, ultimately ensuring a more consistent QoE for all users.
In standard ABR distribution systems, when monitoring indicates a possible video degradation, it's typically because of resource capacity limitations. There's little that network operators can do about it since they don't manage the different end-devices streaming from their network. Placing the network bitrate distribution on the server side gives operators an opportunity to act and resolve the problem before it significantly impacts QoE.
When all video sessions cannot be served at the best available quality, for example during peak hours, operators can gracefully and dynamically preserve the bitrate of higher value content such as live events using server-side stream selection.
Another great benefit is that all users are served consistently, according to the policies defined by the operator as opposed to the quality of implementation of one client against another. This means that players don't have to compete between each other for resources and that the operator can guarantee a constant and guaranteed quality of experience for all clients.
Finally, server-side bandwidth measurements use lower-layer TCP information provided by the server's operating system, they don't suffer the same limitations that clients have when the operators is reducing the network latency and they still work perfectly when video segments are divided into much smaller chunks. As a result, the latency on the network can safely be reduced without taking the risk of provoking frequent rebuffering on players. End users can enjoy live streams that are no longer tens of second late compared with traditional broadcast.
One of the main reasons ABR streaming has been successful is its adaptability to any kind of device and to different types and qualities of connections. In that context, the principle of only giving the client the choice of which resolution and bitrate to use whenever it initiates a video streaming session may have seemed logical so far. Today, the experience shows that it also leads to a certain number of limitations on how network resources are used and how the different devices share these resources between each other. These limitations can, however, be lifted by not relying only on clients' decisions but by also incorporating information from the network.
Server-side segment selection solves an increasingly critical and very real-world challenge for operators, allowing them to deliver a consistent experience across devices and players while optimizing the usage of network resources. This represents an inflection point in the industry, giving operators full control over how resources are used so that they can deliver high-quality video streams, in particular with low latency.
[Editor's note: This is a contributed article from Broadpeak. Streaming Media accepts vendor bylines based solely on their value to our readers.]
In the COVID-19 era, the right approach can check both boxes, and decisions made in response to the current demand on bandwidth can have long-term business advantages for OTT services
By David Springall, Founder & CTO, Yospace . For years, our industry has been trying to improve the video viewing experience. And for many people, this has meant improving image quality - trying to squeeze in and compress even more pixels to offer better images without congesting the available bandwidth.But it seems that a new parameter is now becoming increasingly important for improving quality of experience. Did you catch what it is?
Artificial intelligence and machine learning, along with deep learning and neural networks, are solving OTT challenges from encoding quality to closed captioning.
Companies and Suppliers Mentioned