Inside the EBU’s Dynamic Media Facility: Why DMF and MXL matter for the next era of sports production

Europe’s major broadcasters are confronting a fundamental challenge: how to deliver more live sport, in more formats, to more platforms, while operating within tighter budgets and increasingly distributed production models.

Cloud, IP, software-defined workflows and AI integrations are no longer “experiments” – they are becoming essential components of major events.

Into this context the European Broadcasting Union (EBU) has conceived a new reference architecture: the Dynamic Media Facility (DMF). It’s a blueprint for building flexible, scalable and vendor-agnostic media infrastructure.

“The goal is an ecosystem where broadcasters mix functions from different suppliers without bespoke integration every time. But for that modularity to work, those functions need a common way to exchange media payloads”

And at the heart of DMF is a new open-source component: the Media eXchange Layer (MXL), designed to move video, audio and timed data efficiently between software-defined media functions.

If you read my articles on the Time Addressable Media Store (TAMS), you’ll recall that TAMS is broadly applicable across a wide range of media applications. DMF/MXL complements that by targeting the high-performance, low latency, real-time domain.

Why the industry needs a new facility model

Sports production has for a long time relied on a familiar formula: specialist hardware, OB trucks, fixed control rooms, and dedicated routing infrastructure. This model is increasingly strained by:

• Rising content expectations: UHD, HDR, multi-language, multi-angle, clips for social, and near-live digital packaging.
• Operational distribution: Remote and distributed production require flexible workflows that can be deployed anywhere.
• Cost and capacity pressures: Buying and refreshing hardware for peak load is inefficient and expensive.
• Platform proliferation: Broadcasters are outputting to linear, OTT, FAST, apps, and social all at once.
• Pace of development and cyber threats: Rapid software evolution is the new normal – updates are expected in weeks not years, and security vulnerabilities require patching within hours.

The industry needs an architecture that can scale up or down, move seamlessly across locations, and rapidly integrate best-of-breed tools – allowing broadcasters to update, secure and iterate their production chains with the same agility as modern software platforms.

This is the space DMF is designed to occupy.

What the EBU means by a “Dynamic Media Facility”

The DMF concept is set out in the EBU’s reference architecture white paper. At its core, DMF reframes a broadcast facility as a cluster of interchangeable, software-defined media functions running on general-purpose compute – on-prem, in a data centre, or in the cloud.

At its heart, DMF treats every part of the production chain as a modular, software-defined function that can be deployed on general-purpose compute, orchestrated dynamically, and scaled up or down as needed.

Instead of fixed hardware paths, the facility becomes a flexible cluster where ingest, replay, graphics, mixing and playout run as interchangeable components across on-prem, edge or cloud environments.

Read more Inside TAMS: How Time-Addressable Media Stores could redefine sports workflows

DMF is deliberately vendor-neutral. The goal is an ecosystem where broadcasters mix functions from different suppliers without bespoke integration every time. But for that modularity to work, those functions need a common way to exchange media payloads.

That’s where MXL enters the architecture.

MXL: The media-exchange fabric beneath it all

The Media eXchange Layer is the key technical innovation. While IP flows (eg ST 2110) still get media in and out, MXL acts as a high-performance shared memory fabric within the DMF processing cluster.

Think of it as a virtual internal routing layer for software.

MXL provides:

• High-performance exchange of video, audio and timed data using shared memory (including in future via RDMA).
• Asynchronous communication between media functions (containers), enabling low-latency hand-off without packetising everything.
• A common data structure for media “grains”, timing and metadata.
• Vendor-neutral interoperability, delivered via an open-source SDK (under the Linux Foundation with EBU stewardship).

In practical terms, a replay module, a graphics engine and a multiviewer can all share media without bespoke integrations, without heavy network overhead, and without locking into one supplier’s internal API – focusing purely on media transport while leaving orchestration to the control plane.

If DMF is the facility, MXL is the cabling.

Why this matters for sports production

Sports is the ultimate stress-test for any media architecture. Live environments demand reliability, scale, speed and the ability to adapt workflows rapidly.

DMF + MXL support these needs directly:

1. Elastic scaling for peak events

Olympics, major tournaments, and high-profile matches often require 3-4 times peak capacity compared to normal operations. Under DMF, additional replay servers, clipping services or multiviewer instances are micro-services that can be spun up and retired on demand.

2. Distributed and remote production

MXL supports workflows where ingest might occur at the venue, graphics in a central hub, and quality control in the cloud – all without tying functions to physical locations.

3. Vendor choice and innovation

Sports broadcasters increasingly want to integrate niche or specialist tools – best-in-class tracking, AI-based highlights, cloud vision mixers – without building custom glue code for each. MXL’s open model is designed to support that future.

Challenges and considerations for broadcasters

DMF and MXL represent a significant shift and adopting them requires more than swapping equipment. Key considerations include:

• Infrastructure readiness: DMF expects modern compute clusters, high-bandwidth, low-latency links within and between compute nodes. Broadcasters will need to evaluate what can run on existing hardware and what requires new investment.
• Redefining media functions: Organisations must break down traditional monolithic systems into discrete containerised functions. This is a cultural as much as a technical shift.
• New operational models: Teams accustomed to fixed control rooms and dedicated hardware will need to adjust to orchestrated, software-based environments. Observability, performance monitoring and incident response look different in a DMF world.
• Standards alignment: DMF intersects with other evolving standards – NMOS, ST 2110, SRT, cloud media APIs like TAMS and frameworks like NBMP. Integrators need to understand how DMF fits into the wider stack.
• Vendor maturity: MXL is open-source, but commercial products with full MXL compatibility are only just starting to emerge. Early adopters will lead the feedback loop into the ecosystem.

Adoption will be incremental, with most broadcasters beginning through pilots and hybrid deployments.

A blueprint for the next decade

The EBU’s Dynamic Media Facility and Media eXchange Layer offer a coherent, open and future-ready framework for sports broadcasters facing rapid change. DMF provides the architectural playbook; MXL provides the technical glue. Together, they enable facilities to become more flexible, more scalable, more modular and more efficient.

Sports production will remain one of the most demanding environments in media. DMF and MXL give broadcasters a way to meet those demands while embracing innovation at their own pace.

The next article in this series will explore vendor adoption and product roadmaps – which manufacturers are integrating MXL, what tools are emerging, and how fast the ecosystem is maturing.

Watch Co-chair of the EBU Dynamic Media Facility initiative Peter Brightwell at SVG Europe’s FutureSPORT event speaking about software-based workflows