Broadcasters have spent years trying to integrate different vendor technologies in their facilities. As the industry has moved closer to software, that struggle has become more pronounced. Currently, media exchange protocols need timing synchronization accuracy that’s beyond most IT hardware, while the reliance on SMPTE ST-2110 and NDI can lead to significant computer resource consumption, latency and vendor lock-in.
However, a groundbreaking initiative led by the European Broadcasting Union (EBU), with key contributions from both broadcasters and vendors, could be about to revolutionize how media is exchanged.
The European Broadcasting Union’s Dynamic Media Facility (DMF) project published its first white paper in 2023, recommending the adoption of an IT-focused approach with best practices that involve developing solutions using a layered architecture and creating a unified infrastructure for media production. The idea of the Media eXchange Layer (MXL) came when Canadian broadcaster CBC began to reflect on the move to software-first cloud architecture for its facilities. CBC’s proposed approach also included a layered model in which each layer has distinct responsibilities. At the base lies the physical IT equipment, moving up to applications and user interfaces at the top. The crucial missing piece, they realized, was a software-centric approach to media exchange.
Compute Heavy
“We didn’t want to use a streaming technology like SMPTE ST 2110, for example, or NDI to exchange video, audio and timed data,” explained Félix Poulin, co-chair of the Dynamic Media Facility Group at the EBU and director, global innovation collaborations, CBC/Radio-Canada. “The idea is not to go outside of the software on one computer; we want applications to share video frames and audio chunks directly within a computer’s memory.”
The concept of exchanging memory between applications isn’t new, but most of the currently available solutions are driven by specific vendors, and some broadcasters are not keen on being tied into one vendor’s way of doing things. MXL, therefore, is a direct response to the need for interoperability, tackling each layer of the DMF reference architecture independently, with media exchange being the obvious first target.
“Using a streaming technology like ST-2110 or NDI requires a lot of computer resources,” Poulin added. “You need to packetize the image and organize it in a certain way, and then you need to serialize it and send each packet one by one. To reconstruct the video and audio, buffering happens, which takes time, so there’s latency building up in a complex system.
“But using direct memory access, like memory sharing, saves all of this work,”
he continued. “You save compute, therefore power, therefore efficiency. You solve latency, because the application will put parts of the video image in its memory as it’s working on it, and then that image is available to be picked up there. There’s no additional process.”
Early experiments have shown latencies of less than 1 millisecond for each transfer of video, a significant improvement over the 20 milliseconds per device typically expected with ST-2110.
Why Broadcasters Are Embracing MXL
For broadcasters, MXL is crucial in future-proofing their operations. CBC is currently working on a major project for its Toronto headquarters, Poulin said, aiming for a dynamic and adaptable technology platform. “If we were going to be a static, purpose-built facility like before, with studios that are a certain size and have a certain number of cameras, we would be limited if we decided two years later to do another kind of show with that facility,” he said.
The industry’s shift to software, alongside the issue of getting different vendors’ technology to talk to each other, made it clear that accelerating interoperability was paramount. MXL became the key to “unlock that vendor ‘lock-in’ problem,” Poulin said.
The way we produce live production is changing. The move to software-based workflows is becoming a reality. But software brings its own challenges, such as interoperability and latency. In the past, we could connect any piece of equipment from one manufacturer to another via SDI and it would work reliably and with virtually no latency. As we move to software, we need a solution that is widely supported and works with native software workflows. This is where MXL comes in.
MXL is based on principles already widely adopted in the IT world when memory is exchanged between applications.
Many vendors already have proprietary solutions to share data in memory between their own applications, but the industry needs open ecosystems that enable data to be read and written across multiple tools from multiple vendors.
MXL is being advanced by broadcasters and vendors collaborating via the Linux Foundation, which means that not one single organization or company owns the technology; it is an endeavor shared among those that wish to contribute. That matters in the U.S., where every facility is a best-of-breed mosaic. Producers want their preferred vision mixer here, their favorite graphics engine there and they expect it all to click together regardless of the underlying infrastructure. MXL provides a common, software-centric exchange so those components interoperate cleanly, whether on-prem, in private data centers or in a public cloud.
Importantly, MXL is built on existing principles widely adopted by software and cloud providers and on existing media principles as used in NMOS and ST-2110. It is not revolutionary but evolutionary. By leveraging existing ideas and technologies we can ensure new workflows are already widely supported where we need them to be
There’s a sustainability dividend, too. In the U.S., greener choices accelerate when lower power and rack-space savings translate into lower opex. By avoiding needless serialize/deserialize cycles and enabling smarter placement of compute, MXL reduces CPU burn and cooling overheads, gains that compound at national scale.
MXL will enable U.S. broadcasters to have the reliability they are used to, while gaining agility and maintaining security and control over their assets. The next refresh doesn’t have to be a leap of faith; it can be a confident step into an open, software-first future.
— Ian Wagdin, VP, Technology and Innovation, Appear
MXL will also help broadcasters like BBC, which has resources in multiple locations across the U.K. “Networks are so fast these days, you get hundreds of gigabits per second down inexpensive fiber,” explained Peter Brightwell, lead engineer at BBC research and development and co-chair of the DMF Group. “We have the option of being able to move our resources around the country, to wherever there is the most capacity.
“What we really want is to not just do things in different ways depending on where they are, whether they’re local, within the cloud and so on,” he added. “That’s the idea of a Dynamic Media Facility. About a year ago, we had a workshop in Geneva, then CBC hosted another one in Toronto at the beginning of this year, where we brought together several of those key vendors to challenge the industry to work together. It’s gone very well.”
Open-Source Approach
Instead of the traditional, lengthy standardization process, the EBU and its partners have opted for an open-source approach to MXL. “We decided to go with an open-source approach because usually our industry would have gone through a standardization process at this stage,” Poulin said. “We’d bring in standards bodies, write the text about the solution; eventually we get it ratified and then start to see first implementations.”
Instead, a software development kit (SDK) has been developed, and making it open source means the kit can be used by all vendors and thus become automatically interoperable. “The vendors we approached were enthusiastic and I think most of them were ready to go in that direction,” he said. “The EBU called the first meeting in November to test how far we could go, and everybody in the room said, let’s do it!”
Added Brightwell: “In an open-source project, everybody can see everything that’s happened. And like the best open-source projects, you get to influence it by doing work and actually contributing something to the community.”
But what does this mean for standards such as SMPTE-ST 2110 and NDI? Both Poulin and Brightwell said ST 2110 should continue to play an important role in the industry. “2110 is not going to go away,” Brightwell said. “It’s there at the inputs and the outputs and now the edges.”
Poulin added: “2110 was an enabler to get there. It was the move to get out of industry-specific connectivity like SDI, to go to IP connectivity. It will still be the interface of choice between compute clusters. 2110 will still be used.”
The first release of MXL code is expected towards the end of this year, very likely in the fourth quarter. “In terms of the other layers, we’ve identified a couple of immediate priorities,” Brightwell said. “One is to work on the joined-up timing model. One of the great things about working in software is you can add this sort of thing very easily, the hooks are there to put time stamps into MXL payloads, as they’re called, so that they can be used downstream.
“One of the questions we’re often asked is, if you have two separate pieces of technology, how do they find each other?” he continued. “That’s a fundamental part of what we call the control layer and that is going to be our immediate next step.”
At IBC2025, the EBU announced a partnership with the Advanced Media Workflow Association (AMWA) to create the Joint Taskforce on Dynamic Media Facilities
(JT-DMF), which will bring together vendors, end users and system integrators, and aims to ensure that the next wave of live production infrastructure can benefit users and vendors alike. “There are a lot of questions surrounding orchestration,” Poulin said.
“We need to solve a number of these problems in that category. The joint task force now begins working on these different aspects by priority.”
The momentum behind MXL is undeniable. “It’s been so fast from when we started with the idea, to a number of vendors being willing to work in open source, to releasing an early version of code in June,” Poulin said.
Multivendor demonstrations at IBC showcased the progress, and work is underway to finalize a production-ready Version 1. Poulin hopes that by NAB Show in April, vendors will have solutions incorporating MXL available for purchase.
MXL represents a significant leap forward for the broadcasting industry, promising greater interoperability, efficiency and adaptability in an increasingly software-driven world.
This article originally appeared in TV Tech sister brand TVBEurope.
