The technological innovations of generic IT infrastructure and cloud computing look compelling to address these new media consumption habits through software-only facilities running on-prem or in the cloud. Unlike broadcast setups reliant on clock-driven signal synchronization, IT equipment and cloud systems operate in an event-driven, asynchronous manner. This presents us with the opportunity to build systems that match or exceed today’s capabilities. To do so, we need to fundamentally re-evaluate how we manage live video. However, before looking ahead, we must understand our heritage.

Broadcast workflows, specifically live productions, have stringent, valued, and engineered requirements of low latency, high quality, and redundant systems that are responsive and provide best-of-breed vendor choices. The challenge of moving forward is reconciling these requirements to a generic IT environment – be it on-prem or in the public cloud. Let’s start by looking at the technology stack that enables present-day capabilities.

Current Technology Stack

Today, software-based solutions are primarily built as appliances that connect to larger hardware-driven broadcast facilities. Commodity PCs are at the base of it all. A middle layer, the “plumbing”, provides a software abstraction of the hardware broadcast infrastructure. The plumbing includes drivers for professional IO interfaces (like SDI and ST 2110 NICs) and low-level software processing libraries (for buffer handling, timing, synchronization, packet processing, hardware control, etc.). This middle layer abstracts underlying technological details and allows for domain specialization within applications. We, at Matrox Video, have dedicated a lot of our efforts and time to making and providing this middle layer to the industry with field-proven SDKs, APIs, and professional IO interfaces throughout several technological shifts.

Developers with expertise in a specialized domain provide media software (like graphics, replay, switchers, production servers, master control, etc.) built on this foundational middle layer to resolve and enable broadcast workflows (Figure 1). This middle layer has benefitted both developers and broadcasters alike. For developers, it shelters them from low-level technological changes, allowing them to focus on their core competencies. Meanwhile, broadcast professionals can choose from a variety of domain-specific best-of-breed software applications based on the specialized features they require, without getting bogged down in the details of low-level infrastructure.

Adapting To The Changing Landscape

When faced with the challenge of operating in the cloud, or more specifically a generic IT environment, there is the natural temptation to reuse what we have built so far with a lift-and-shift approach. This generally means running the same monolithic applications as is on a virtual machine in a public cloud. It also means trying to virtually replicate the timing and synchronous connectivity of hardware-based systems. In other words, interconnect monolithic stacks with a synchronous software interconnect (rather than a synchronous hardware interconnect). This approach brings limited gains. Scalability quickly becomes a problem and compromises on the broadcast principles, of latency, quality, and frame-accurate deterministic control– all within a best-of-breed environment–invalidate this as a viable solution at scale.

As we look at using IT-only infrastructure on-prem or in the cloud, the current clock-driven, synchronous technology stack is no longer suited to meet the requirements of broadcasters. The industry’s reliance on single-purpose hardware and software, interconnected through real-time interfaces like SDI and ST 2110, is incompatible with cloud technologies.

In a software-defined infrastructure, a single PC is replaced by distributed computing that can be spun up or down on demand, in a public or private cloud. As mentioned, broadcast infrastructure is built around hardware-driven synchronous interconnects, where operations occur in a predetermined, time-aligned linear sequence – locked to one master clock. This translates to a lock-stepped, frame-by-frame progression of media processing and transport.

The cloud, specifically generic IT infrastructure, is asynchronous where processing and transport will be event-driven. IT architectures allow the rapid transport and processing of time-sensitive data and as a result, IT architectures can swiftly transport and process such data without the need to serially connect with clocked interfaces. In other words, processing and transport happen as fast and as soon as possible.

Re-imagining The Technology Stack

To truly achieve the objectives, a fundamental rethink of how we design and deploy broadcast systems is needed. In this rethink, we have the chance to use IT equipment the same way as other industries and exceed the performance we attain today in hardware-based facilities. The key is to embrace the asynchronous nature of IT and forgo our synchronous heritage for everything except for mandatory real-time interconnects to “glass” (i.e., cameras, monitors, or projections onto these).

In this new paradigm, the technology stack starts at the compute level – CPUs, GPUs, networking, and storage – whether they be public cloud instances or on-prem servers with professional SDI cards or ST2110 NICs. The next layer is a software framework that takes care of the resilience, time coherence, flow, and scalability of uncompressed media. The media processing is done by stateless media services that run on top of this framework and can be distributed across many systems. These media services can be built using simple APIs – examples include graphics rendering, mixers, audio processing, scalers, etc. The application control is done through web APIs to provision and control the media services. Finally, exhaustive metrics allow for observability in the framework to orchestrate, automate and maintain large-scale systems (Figure 2).

This technology stack provides the benefits of scalability, resilience, agility, and composability of an IT environment to software broadcast facilities. With empirical tests, we see that this new technology stack not only meets but exceeds the performance of current practices – making software-centric facilities even better than the hardware-based facilities that we set out to match.

Our Reinforced Commitment To The Plumbing

In this new paradigm, we continue to strive for the same goals we have always had in providing the plumbing to the industry. We introduced an asynchronous media framework as the newly re-imagined middle layer of the technology stack (i.e., the software broadcast infrastructure). This middle layer provides the benefits of IT for developers and broadcast professionals, enabling focus on core competencies and offering flexibility in production without compromising requirements.

In an IT-based broadcast environment, the middle layer of the technology stack we discussed becomes increasingly critical, as it is now the core broadcast infrastructure. Our commitment to serving as an intermediary in the broadcast ecosystem extends beyond maintaining technological stability; it is about actively shaping the broadcasting ecosystem to be more adaptable, resilient, and innovative.

With our asynchronous media framework Matrox ORIGIN, agile software broadcast facilities with customized workflows can be dynamically spun up with a choice of media services interconnected, ensuring guaranteed synchronization for robust and reliable live production workflows with necessary redundancy. This software infrastructure enables scalability and customization, accommodating both small-scale productions with a single commentator/operator and large-scale dynamic deployments with multi-operator gallery control or automation. In these agile software broadcast facilities built on Matrox ORIGIN, production possibilities are no longer hindered by the technology stack, enabling decisions to be driven solely by business opportunities.