Mathieu Rochon, Senior Technical Advisor at CBC/Radio-Canada, outlines for TM BROADCAST the principles behind the corporation’s Universal Production Platform and reflects on the transition to IP, software-defined workflows, and the role of MXL in next-generation media infrastructures

*This content is included in the July issue of TM BROADCAST. You can access the full magazine here.

The conversation in the broadcast industry is gradually shifting from the adoption of SMPTE ST 2110 towards a new layer of abstraction, represented by concepts such as the so-called media exchange layers. In this context, approaches such as DMF (Dynamic Media Facility), developed by the EBU (European Broadcasting Union), and MXL (Media Exchange Layer) are gaining traction, particularly among the most advanced players in the sector.

To understand how this shift is being addressed in practice, we turn to one of the most advanced cases in this field: CBC/Radio-Canada. Together with Mathieu Rochon, Senior Technical Advisor at the corporation, we explore how these concepts are being implemented in a real-world environment, shaped by the rise of streaming, OTT, and the need to rethink production models.

As Mathieu explains, a commitment to technological innovation is part of the organisation’s DNA: “Since the early 1990 with one of the first large scale SDI production center in Toronto to our recent fully 2110 multi-platform production facility in Montreal, we want to continue to be in a position to influence the future of our industry, and there is no better spot to do so than to be at the forefront of the technological evolution.”

The Universal Production Platform (UPP) developed by CBC/Radio-Canada is closely aligned with the Dynamic Media Facility (DMF) architecture defined by the EBU, where CBC/Radio-Canada has recently become a Full Member after 75 years as an Associate Member

Universal Production Platform (UPP): four key principles

The Universal Production Platform (UPP) developed by CBC/Radio-Canada is closely aligned with the Dynamic Media Facility (DMF) architecture defined by the EBU, where CBC/Radio-Canada has recently become a Full Member after 75 years as an Associate Member. The platform is structured around a set of principles aimed at redefining how production environments are designed and operated.

At its core, Mathieu explains, the system is built on four fundamental pillars.

Principle 1: A clear separation between Production Areas, Operator Positions and Shared Production Resources.

• If it’s required to capture or create sound/images, it goes in production areas.
• If it’s required to monitor and control signals/equipment it goes in operator positions.
• If it alters/modifies/processes any audio/video signal it goes into the pool of shared production resources.
• Separation of the back end (media functions) and the front end (user interface)

Principle2: Common routing infrastructure based on routable IP technologies and a limited number of protocols / formats/ codecs.

• We will no longer be restricted to a fixed size router, literally everything will be routable using the IP network.
• The router becomes agnostic to production formats (resolution, frame rate, aspect ratio, etc.). But support for 1080p59.94 10 bits is mandatory on day 1.
• Using the proper codecs, IP can be extended anywhere, including the public cloud and your own home office.

Principle 3: Conscious bias towards software based solutions running on COTS hardware.

• A common server infrastructure can run multiple different applications.
• One server can serve as a multiviewer one day and a replay server the next.
• Allows the use of public cloud to temporarily augment processing capacity.
• Ability to rent additional licences to meet extra production needs.

Principle 4: No fixed and permanently assigned or reserved production resources.

• Dynamic allocation of production resources between multiple facilities.
• Ability to right size infrastructure based on true simultaneous production needs.
• Resources can be reserved through scheduling and management tools.
• Deployment, configuration and updates are facilitated by standard IT automation tools.

“We need to stop thinking in terms of ‘video formats’ (like 4K or HDR) and start thinking in ‘form factors.’ A show, five years from now, is no longer just a linear program; it’s a content ecosystem. It’s a documentary that is also a vertical social series, an interactive community hub, and a searchable archive”

What needs does it address?

The Universal Production Platform (UPP) at CBC/Radio-Canada responds to a clearly defined set of operational and strategic needs, within the context of a large-scale technology renewal process.

As Mathieu Rochon explains, the starting point was highly pragmatic: “In Toronto, the heart of our UPP project, much of the technology and infrastructure is at the end of life, so we are bound to do something no matter what.”

This necessary renewal became an opportunity to rethink workflows and production models from the ground up. In this context, the project addresses both operational challenges and broader strategic objectives.

Operational benefits

From an operational standpoint, one of the key priorities is the simplification of the user experience. “Our project is closely tied on the EBU Dynamic Media Facility (DMF) reference architecture and a unified user interface independent of the underlying software media function is at the heart of our design. In the early implementation of Universal Production Platform (UPP) that we used for the Olympic Winter Games Milano Cortina 2026, we in-house developed our own user interface that runs in browser on laptop.”

Another key pillar is the enablement of remote workflows. “By choosing the right protocol and codec, webRTC, opus, H.264, VP9 or AV1, we can have a remote user experience with low enough latency that it is not only possible to do linear production workflow but also production workflow where feedback is involved like a double ender.”

At the same time, the platform enables a rethinking of how resources are shared, in direct alignment with the dynamic allocation principles embedded in the UPP architecture.

“By decoupling hardware from specific rooms, we aim to make our studios, control rooms, and edit suites as frictionless to book and deploy as a standard conference room”

Strategic benefits

From a strategic perspective, one of the primary objectives is the optimisation of both technological infrastructure and physical spaces. “Embracing the DMF architecture, moving from dedicated hardware to software running on COTS supported by a realtime media IP network (2110), we will reduce equipment room footprint and will enable production usage utilization.”

This approach makes it possible to transform studios and control rooms into dynamically assignable resources. “By decoupling hardware from specific rooms, we aim to make our studios, control rooms, and edit suites as frictionless to book and deploy as a standard conference room. By centralizing compute power, it allows us to deliver high-end production tools to regional bureaus via Thin UIs. This eliminates on-site infrastructure costs while strengthening the local coverage core to our Canadian audience.”

Another key strategic driver is the ambition to maintain a leading position in the adoption of emerging technologies. As Mathieu notes, “CBC/Radio-Canada has a long history of being at the edge of technology. Since the early 1990 with one of the first large scale SDI production center in Toronto to our recent fully 2110 multi-platform production facility in Montreal, we want to continue to be in a position to influence the future of our industry, and there is no better spot to do so than to be at the forefront of the technological evolution.”

Finally, the platform is conceived with a long-term vision aligned with evolving consumption patterns. “With the media landscape dramatically shifting toward OTT platforms and social media in the last few years and with no sign of slowing down, we need to give CBC/Radio-Canada the most scalable and the most flexible system possible in order to be able to rapidly pivot our production requirements and output formats.”

Software-defined production: towards dynamic orchestration

From a workflow perspective, the transition towards DMF-based architectures is reshaping how content is produced, managed, and exchanged at CBC/Radio-Canada. “We are shifting from static configurations to dynamic orchestration, utilizing repositories to manage infrastructure state and CI/CD pipelines for resource deployment”, explains Mathieu.

At the heart of this transformation lies a key concept: the abstraction layer. “We are investing in internal development to glue diverse software solutions into a unified UI. Our strategy prioritizes intuitive tool grouping over a ‘one-size-fits-all’ approach. Built on operator feedback, this common layout will enable people to transition between productions with minimal retraining. This allows us to maintain a friendly UX for production staff while the engineering team leverages a modern DevOps stack to manage content exchange and resource allocation.”

At the heart of this transformation lies a key concept: the abstraction layer. “We are investing in internal development to glue diverse software solutions into a unified UI”

MXL: when applications ‘speak’ to each other

Within this broader transformation, the concept of the Media Exchange Layer (MXL) is emerging as a key enabler of interoperability between systems and applications. Mathieu Rochon describes it as a fundamental shift away from traditional signal-based workflows: “The Media Exchange Layer (MXL) acts as a high-performance common media fabric that allows disparate media applications to ‘speak’ to each other directly in memory.”

In contrast to conventional architectures—where media needs to be packaged into streams for transport—MXL eliminates the need for these intermediate steps: “In traditional setups, software vendors are forced to mimic hardware behavior—packaging media into streams (like ST 2110 or NDI) just to move it from one application to another. MXL replaces these ‘virtual cables’ with a shared memory architecture, allowing multiple media functions to access the same raw data simultaneously without constant re-encoding or streaming overhead.”

This approach delivers several key advantages. First, improved computational efficiency: “By eliminating the need to generate, encode, and buffer streams at every ‘hop,’ CPU and GPU resources are freed up to focus on actual media processing rather than just data transport.”

Second, near-zero latency. “Streaming protocols require buffers and handshakes,” notes Mathieu. “MXL allows for instantaneous handoff between functions (e.g., from a vision mixer to an encoder), which is critical for live, real-time broadcast environments. In our early testing with FFMPEG using MXL to encode video to deliver it to our UI, we were able to reduce the latency from 1800ms down to 250ms. This gain is not solely due to MXL, we were able to reduce the encode time and it also simplified the pipeline, but a good deal of it is due to MXL.”

This is complemented by true multi-vendor interoperability. “It breaks the ‘vendor silo.’ Because MXL provides a standardized way to access media in memory, broadcasters can build a “best-of-breed” infrastructure where software from different companies works together as if it were a single, unified application.”

Finally, MXL introduces a highly scalable model. “It enables a ‘compute node’ philosophy. You can scale your processing power horizontally across multiple nodes without the added latency that streaming would add.”

In essence, the shift is “from interconnecting separate boxes to integrating a software ecosystem”, as Mathieu summarises: “It trades the inefficiency of internal streaming for the speed of shared memory, delivering latency close to hardware with the flexibility of the cloud.”

In this context, a key question arises: to what extent are initiatives such as MXL and DMF redefining traditional television production models?

For Mathieu, the answer is clear: “MXL and DMF are not the ‘why’ of change—they are the ‘how.’ While market forces and shifting consumption habits (like the move to social-first content) are the true disruptors, MXL and DMF provide the unified, reconfigurable backbone needed to respond. They don’t redefine the production model themselves; they grant us the strategic flexibility to redefine it whenever the market shifts.”

SMPTE ST 2110: beyond the technological transition

Alongside these forward-looking initiatives, CBC/Radio-Canada has already undergone a significant transformation with the implementation of a full SMPTE ST 2110 facility. An experience that, as Mathieu points out, went far beyond a purely technical transition. He played a central role in the project: “My specific focus was on broadcast controller integration and workflow development, and it’s from that vantage point that I saw our biggest hurdles.”

From his perspective, the shift to IP is as much cultural as it is technological: “The transition to ST 2110 is as much about a cultural shift as it is a technical one.” On the technical side, one of the main challenges lies in the scale and complexity of deployment: “The sheer scale of configuration is easy to underestimate; you’re looking at a 2x to 5x increase in multidisciplinary manpower compared to traditional baseband.”

In this context, automation becomes essential rather than optional: “To survive that, automation is mandatory. We had to embrace a DevOps toolkit—Arista AVD, Ansible, and Jenkins—to manage our network infrastructure.” He adds: “Most importantly, we learned that a spreadsheet is not a viable way to manage IPs in a large-scale facility. Investing early in a ‘Single Source of Truth’ like Netbox is critical for long-term stability.”

Beyond the technical layer, however, organisational change proved equally decisive. “The challenge wasn’t just ‘Broadcast vs. Network’ skills”, explains Mathieu. “We found that success doesn’t depend on your background, but on a thirst to learn. To manage the transition, we performed a deep skill-gap analysis across our project and support teams. We focused on training a ‘cohort of champions’ who could then peer-coach the wider group.”

Was it worth the effort? For Mathieu, there is no doubt: “Absolutely! Our operators have no desire to ever go back to baseband. The need to learn and innovate with 2110 has also put us in a much better position to tackle an implementation of a DMF facility in Toronto with our Universal Production Platform.”

“While market forces and shifting consumption habits (like the move to social-first content) are the true disruptors, MXL and DMF provide the unified, reconfigurable backbone needed to respond”

“Moving towards IP has had the biggest tangible impact so far”

Looking at the broader picture, the move towards IP has already had the most tangible impact on the organisation in recent years. “Moving towards IP has had the biggest tangible impact so far. By overcoming the physical constraints of baseband routing, we unlocked the ability to pool and manage resources regardless of proximity to the router”, explains Mathieu.

This shift has also aligned the broadcast sector more closely with the wider IT ecosystem: “This pivot encouraged the broadcast industry to leverage proven IT innovations rather than proprietary hardware. That said, the transition to ST 2110 was the catalyst for the industry’s shift toward software-defined productions. Ultimately, this software-centric model democratizes access to high-end tools for our creative teams. This will be the next big technology wave.”

Looking ahead: a five-year outlook

Looking ahead, Mathieu Rochon sees the evolution of public broadcasting as closely tied to the adoption of DMF-based architectures. “It isn’t just about choosing one technology—like AI or Cloud—it’s about creating a unified, containerized environment that can support all of them simultaneously.”

In this context, he identifies three key pillars that will shape the next phase of development. The first is technical agility (the foundation): “We are moving away from fixed hardware toward a ‘Media-as-a-Service’ approach. By adopting a DMF model, we treat production resources as a fluid pool of compute power. This allows us to scale up for a major event like the Olympics and scale down the next day, regardless of whether that processing happens on-prem, in the cloud, or at the edge.”

The second pillar is editorial distinctiveness (the core). “Public Service Media (PSM) must double down on what we do best: relevance, accuracy, and trust. In an era of ‘AI-generated slop’ and fragmented truth, our value lies in high-quality, human-led journalism and local cultural creativity. Technology’s job is to make it so our creators can focus entirely on the editorial and creative narrative.”

Finally, Mathieu points to a transformation in the very conception of content: the evolution of the ‘show’ (the output). “We need to stop thinking in terms of ‘video formats’ (like 4K or HDR) and start thinking in ‘form factors.’ A show, five years from now, is no longer just a linear program; it’s a content ecosystem. It’s a documentary that is also a vertical social series, an interactive community hub, and a searchable archive.”

Taken together, these elements point towards a fundamental shift in identity: “The DMF model allows us to stop being ‘broadcasters’ in the traditional sense and start being dynamic media platforms that can meet the audience wherever they are, without reinventing our infrastructure every time a consumer trend emerges.”

“In the early implementation of Universal Production Platform (UPP) that we used for the Olympic Winter Games Milano Cortina 2026, we in-house developed our own user interface that runs in browser on laptop”

The role of public service media

Within this broader transformation, Mathieu also highlights the distinctive role that public service media should play within the industry. “Because Public Service Media is sustained by public funding, we are in a unique position to act as an engine for the entire industry’s health. Unlike commercial players who may seek proprietary advantages, our role is to foster a transparent, interoperable ecosystem. By championing open-source initiatives like MXL, we aren’t just building a tool for ourselves—we are creating a ‘common good’ for the broadcast community. This collaborative approach lowers the barrier to entry for smaller creators and vendors, promotes healthy competition among vendors, and ensures that the future of media remains open, flexible, and sustainable for everyone.”

“Ultimately, this software-centric model democratizes access to high-end tools for our creative teams. This will be the next big technology wave”