Does Eco-Friendly Streaming or Green Streaming Truly Exist?

Jean-Marc Jancovici's answer

Can we consider live 'retranscoding' as green streaming?

When discussing the concept of "green streaming," it is important to consider the environmental impact of various streaming methods. Live retranscoding involves converting a video stream from one format or bitrate to another in real time to ensure compatibility with different devices and network conditions. While this process can improve viewer experience, it raises the question: can we consider live retranscoding as green streaming?

"Just-in-time" transcoding is often touted as a beneficial feature, as is 'Serverless', even by companies that claim to prioritize eco-friendly operations. Essentially, "just-in-time" transcoding means that a source file already exists (quite often already in a ready-to-broadcast format), but the publisher (as suggested by the transcoding provider) may decide to transcode it again, usually for peculiar reasons, such as adding easily bypassed Digital Rights Management (DRM). Real-time transcoding is the least energy-efficient method and may be used for content that has already been transcoded into the final output format previously (as reruns are common).

First, let's look at the energy consumption associated with live retranscoding. This process often requires dedicated hardware or cloud-based servers to handle the computational workload, resulting in significant energy usage. Additionally, live retranscoding can generate heat, which may necessitate cooling systems that consume even more energy.

Second, live retranscoding can contribute to increased network traffic. As multiple versions of the same content are generated to cater to different devices and connection speeds, more data needs to be transmitted across the internet, leading to higher energy consumption in data centers and network equipment.

Finally, the e-waste generated by the hardware used for live retranscoding should not be overlooked. As technology advances and hardware components become outdated, they are often discarded, contributing to the growing problem of electronic waste.

Why do we need just-in-time transcoding?

Some not-said answers could be 'to charge more' as this is sometimes not necessary at all but still done. Just-in-time transcoding may not be necessary for free-to-air content because most devices support the HLS (HTTP Live Streaming) standard, even if not in its latest version. Since HLS is a widely adopted streaming protocol, compatibility issues are minimized, and there is no need to transcode the content in real-time for various formats or codecs. Using a single standard like HLS simplifies streaming and reduces the energy and computational resources required for transcoding.

However, just-in-time transcoding might be needed for DRM (Digital Rights Management)-protected streams because not all devices support the same DRM systems. Different devices and platforms may use various DRM technologies, such as Widevine, PlayReady, or FairPlay. To ensure compatibility and maintain content protection, it may be necessary to transcode the content in real-time to accommodate the specific DRM requirements of each device or platform. This real-time transcoding process ensures that the content is (somewhat) protected while also being accessible across a wide range of devices.

What is the most energy-efficient codec that also offers the best device compatibility?

The H.264 codec (also known as AVC, or Advanced Video Coding) is currently the most widely adopted codec that strikes a balance between energy consumption and device compatibility. H.264 offers good video quality at relatively low bitrates and has been the industry standard for many years. As a result, a large number of devices, including smartphones, tablets, computers, and smart TVs, support H.264 video playback.

While newer codecs like H.265 (HEVC) and AV1 offer improved compression efficiency, which can lead to energy savings during content distribution and playback, they may not be as widely supported by devices yet, especially older ones. Additionally, these codecs can be more computationally intensive during the encoding process, which may result in higher energy consumption compared to H.264.

How much more energy does AV1 consume compare to H264 for encoding ?

The energy consumption of video encoding depends on the specific codec and encoding settings used. AV1 is a more modern and efficient video codec compared to H.264, providing better video quality at lower bitrates. However, AV1's improved compression efficiency comes at the cost of higher computational complexity during encoding.

While it's difficult to provide an exact figure for the energy consumption difference between AV1 and H.264 encoding, it's generally accepted that AV1 encoding can be significantly more computationally intensive than H.264 encoding. Depending on the specific encoding settings, AV1 may consume anywhere from 5 to 20 times more computational resources than H.264 for encoding the same video content.

This increased computational demand leads to higher energy consumption during the encoding process. However, it's important to consider that the more efficient compression offered by AV1 results in smaller file sizes and lower streaming bandwidth requirements. This can lead to energy savings during content distribution and playback, offsetting some of the increased energy consumption during encoding.

Is 'serverless' purely a marketing thing or does it actually exist ?

"Serverless" is not purely a marketing term; it refers to a genuine concept and architectural approach in cloud computing. Serverless computing allows developers to build and run applications without having to manage the underlying infrastructure. In a serverless architecture, cloud service providers dynamically allocate resources and automatically handle infrastructure management tasks, such as server provisioning, scaling, and maintenance.

The term "serverless" can be somewhat misleading because it may imply that there are no servers involved. In reality, servers are still present, but developers don't have to worry about them as the cloud provider takes care of the infrastructure. Serverless computing often operates on a pay-as-you-go model, where users are billed based on the actual compute resources consumed, rather than pre-allocated capacity.

Serverless architectures often rely on Functions as a Service (FaaS) platforms, such as AWS Lambda, Google Cloud Functions, or Microsoft Azure Functions. These platforms enable developers to write and deploy individual functions that are triggered by specific events or requests. This approach allows for greater flexibility, faster development, and potential cost optimization (when using them only on-demand), as resources are only consumed when the functions are executed.

Does serverless make sense for 24/7 streaming operations?

While serverless computing offers several benefits, such as scalability, faster development, and reduced infrastructure management, it may not be the most suitable choice for 24/7 operations like TV channels, particularly in terms of cost and energy consumption.

Serverless architectures typically operate on a pay-as-you-go model, where users are billed based on the actual compute resources consumed. For sporadic or variable workloads, this model can be cost-effective. However, for continuous 24/7 operations, costs can quickly accumulate due to the constant use of compute resources. In such scenarios, traditional server-based architectures or virtual machines with fixed pricing models might be more cost-effective.

In terms of energy consumption, the on-demand nature of serverless computing can lead to more efficient resource utilization for variable workloads, but this advantage is less relevant for 24/7 operations. In fact, dedicated servers or virtual machines that are optimized for continuous workloads may consume less energy overall.

Some green transcoding servers that a few broadcast professionals will recognize

Another consideration is that serverless platforms might not provide the same level of customization, control, and performance tuning as dedicated servers or virtual machines. This can be crucial for TV channels that require specific configurations or optimizations to deliver high-quality, uninterrupted streaming.

Being Greener should be a target, rather than spending time greenwashing existing practices in PR battles

In light of these factors, it becomes evident that live retranscoding, just-in-time transcoding, and serverless operations cannot be considered green streaming. Instead, more sustainable alternatives should be explored, such as adaptive bitrate streaming storage for reruns and other 'live-like' broadcasts (VOD2Live), efficient video codecs (also for energy consumption), and edge caching. These technologies can help reduce energy consumption, network traffic, and e-waste while still providing a high-quality streaming experience for viewers.

For that reason, the TV channel demoed below is likely 'greener' than most initiatives currently labeled as 'green' in the streaming industry (especially for companies advertising their "green" focus at all trade shows worldwide, often involving 10 or more people traveling). Not only does it utilize the most energy-efficient CPU for transcoding, but it also avoids retranscoding assets that have already been transcoded. This sets it apart from many advertised green services that fail to implement either of these practices.