Red Hat Device Edge for Industrial Applications: A Journey from Datacenter to Plant Floor

For the last 30 years, Red Hat Enterprise Linux (RHEL) has been the solid foundation of datacenter and cloud infrastructures, powering everything from websites, databases, and applications. While there’s been huge adoption in these areas, Red Hat asked a question about two years ago: Could the same rock solid operating system also be used for industrial applications, and achieve the same dependability and performance demanded by the market?

This blog is a look through the progress made over the last few years, the advancements and improvements, and a glimpse into the roadmap of Red Hat Device Edge for industrial deployments.

Why the desire for modernization

Although industrial systems have evolved, they often still face significant challenges in integration and are not designed to easily adopt new features via seamless, in-place upgrades. This limitation can hinder adaptability and modernization efforts, making it harder for organizations to update or expand functionality without substantial changes to the underlying infrastructure. This inflexibility in handling emerging workloads like computer vision, generative AI, or data aggregation, can stifle innovation and present real challenges for businesses looking to modernize.

Enter Red Hat Device Edge: a modern, flexible foundation that brings all the benefits of open source, as well as Red Hat’s expertise in enterprise software to the far edge. Red Hat Device Edge provides modern functionality for running current and next-generation workloads, all on a single, unified platform.

Red Hat's first wave of expanding out of traditional data centers into industrial applications happened a few years earlier with modern manufacturing execution systems and manufacturing operations management systems implemented at the edge, in the factory and closer to where data is generated. One example here is Red Hat’s partner German Edge Cloud. Their ONCITE Digital Production System (DPS) product is delivered to the shop floor to enable track and trace, serialize or energy monitoring on top of Red Hat OpenShift, providing more transparency into the production process. The next logical step was to close the loop and also look at control systems, acknowledging their even stronger requirements around availability, cybersecurity, time-sensitive networks and constrained hardware - and all at a much larger scale in numbers.

In the beginning

Development of Red Hat Device Edge began as customers and partners began asking: could Linux run at the far edge, well outside the air-conditioned walls of a datacenter or cloud? And if it could, what additional tooling and technologies needed to be developed and included to create an easy-to-consume, nearly perfectly reliable, and highly deterministic version of Linux?

To meet these requirements, Red Hat Device Edge began as a sort of “offshoot” from RHEL, using all the same bits and bytes as RHEL, but assembled through a different mechanism. Instead of hundreds or thousands of individual packages assembled to form an operating system, a new approach was taken: creating point-in-time images, composed by summarizing individual packages, which became the edge-focused operating system base of Red Hat Device Edge.

This approach met a few key requirements from customers and partners:

• Images are immutable, preventing deployment errors or undesired changes to the deployed systems over time
• Updates are delivered as deltas, requiring very little bandwidth to update
• Because each image is individually bootable, automatic health checks and rollbacks became standard features
• A slimmed down base package set provided a compact, portable base for lower-power devices
• Full feature parity with Red Hat Enterprise Linux including customization of the desired image that ensures your OS is fit for purpose, for whatever you need

The solution bridges the gap between IT systems and OT on the shop floor, requiring little to no onsite IT or Linux expertise for initial deployment or ongoing maintenance. This translates into significant time and cost savings, as you no longer need to dispatch specialized experts to remote locations like wind farms, oil rigs, or factories where OT staff may have limited IT knowledge.

By leveraging open software, which is already well-established in the IT landscape, the solution helps provide access to a wider pool of skilled professionals and minimizes the risk of vendor lock-in with costly, proprietary technologies.

Automated and standardized update processes reduce the likelihood of human error during manual updates, while rollback capabilities provide a safety net, minimizing downtime. This makes it possible to schedule updates within short maintenance windows—sometimes as brief as a factory coffee break. More frequent updates also help verify that systems are regularly patched with the latest security updates, significantly reducing the risk of cyberattacks.

Red Hat Device Edge started out with all this functionality, providing a highly dependable offering, but has evolved even further over time.

The first step into industrial - software control

With a firm base in Red Hat Device Edge, the next step was to start testing for industrial workloads. The first main showcase of this was running a software-based bumpless control runtime, provided by Schneider Electric, paired with an interoperable control system jointly developed with Intel to control a sample water tank system: a scaled down representation of waste-water treatment. This demo featured small form-factor systems, running Red Hat Device Edge, which ran the software control across a bank of devices, seamlessly failing over when an individual device failed, or was taken out of service.

This implementation showed that Red Hat Device Edge could be leveraged to develop new functionalities at the edge and re-use already developed technologies in the Red Hat stack to serve the industrial requirements.

As a follow-up to this initial showcase, earlier this year, an update for the state of the art industrial control systems was published with Universal Automation, Schneider Electric, and ExxonMobil, that highlighted how next-generation control systems, powered by modern platforms such as Red Hat Device Edge, are becoming a reality in the industry.

Next up - deterministic performance

With momentum building, a next development milestone was set: bringing the deterministic performance of the platform in-line with other solutions available in the industrial market today. It isn’t enough to simply have low response times, industrial applications, specifically software defined control, require consistent predictable performance - called deterministic performance - when operating, to deliver consistent cycle times and reliability.

Red Hat has been contributing to and supporting the real time Linux kernel for years with Red Hat Enterprise Linux, enabling  a far more deterministic behavior and resulting in much more predictable latency:

In this figure, the blue represents the response times of the standard kernel without the real time capabilities, while the green band shows the far more predictable response times of the real time kernel with the real time scheduler and tuning.

In addition, work has been underway with silicon partners, such as Intel, to implement and enable additional capabilities between hardware and software, further improving the deterministic performance of Red Hat Device Edge. These advancements have included enabling of cache allocation technology, as well as Intel® Cache Allocation Technology (Intel® CAT) and Intel® Time-Coordinated Computing (Intel® TCC) for time-sensitive networking, to allow for less jitter when running real-time tuning and optimizations with Intel’s Caterpillar tool for far more deterministic performance when communicating over a network stack to network targets, such as remote IO, device gateways, PLCs, and more. This work has culminated in impressive results, highlighted by joint benchmarks that use real-world scenarios to highlight the performance of the platform.

In addition, as with all development of Red Hat software, the engineering developments and code have been contributed upstream in the respective open source communities. Through mentorship and leadership in the open source communities, the real time capability has officially been merged to the mainline Linux kernel tree.

Improving the image build process

This year at Red Hat Summit 2024, Image Mode for Red Hat Enterprise Linux was announced. This new shift towards “bootable containers” significantly simplifies the build process for images by leveraging the same syntax and flow as building any other container images, allowing for easier integration with existing toolchains and development flows. In addition, images now become far more portable, as they can be deployed to a containerization platform, such as Red Hat OpenShift, as a virtual machine (VM) on virtualization platforms, or to bare metal systems, such as small form-factor industrial PCs. Due to all these deployments sharing the same image, a higher degree of consistency is achieved, allowing for true “write once, deploy anywhere” development, including all the way out to the far edge.

Adding in lightweight Kubernetes functionality

Another aspect of  Red Hat Device Edge has been the development of a lightweight Kubernetes layer, known as Microshift, which provides container orchestration functionality on Red Hat Device Edge, while maintaining the focus on small form-factor devices. This additional functionality provides the ability to leverage Kubernetes-native functionality and tooling, allowing for modern application development and deployment all the way out to the far edge.

This functionality has been adopted by ABB, who are using Red Hat Device Edge as the foundation for ABB Ability Edgenius, a digital enablement product used to streamline operations and perform advanced data aggregation at the edge.

Supporting the platform through automation

While improvements to the core platform offering of Red Hat Device Edge are essential, a supporting element has been evolving as well: management and automation functionality, provided by Red Hat Ansible Automation Platform. The core platform has received significant improvements to scaling through the use of automation mesh, allowing for support for tens of thousands of devices deployed to the far edge. Also, new certified content has been introduced to simplify the overall management experience.

In addition, Ansible itself has been gaining additional functionality to manage industrial devices over their native connectivity protocols. Red Hat and Rockwell Automation collaborated on connectivity functionality for Ansible, allowing it to begin managing industrial devices, such as PLCs, identically to how other assets such as virtual machines and network switches are managed.

Rockwell Automation’s confidence in Red Hat Ansible Automation Platform has also extended to their core distributed control system, overhauling and modernizing the deployment and lifecycle management of their software stack, something their customers are eager to adopt.

New architectures for new workloads

On top of a growing catalog of certified edge systems from companies like Advantech, OnLogic, Siemens and Welotec, with industrial PCs and fanless systems designed for harsh environments, Red Hat Device Edge has also grown to support different compute architectures to enable next generation workloads, such as predictive AI and generative AI. Red Hat Device Edge is now available for the NVIDIA Jetson Orin and  NVIDIA IGX Orin systems in tech preview. With NVIDIA IGX Orin, Red Hat Device Edge is available coupled with the NVIDIA AI Enterprise software platform for seamless IT and OT convergence. Now, containerized AI workloads can be deployed to Red Hat Device Edge systems, and fully utilize the GPU functionality of the systems for sensor processing, video analytics, functional safety, and other industrial workloads. Standard interfaces leverage USB cameras, drive displays, and more.

As next-generation workloads migrate further out to the far edge, Red Hat Device Edge will continue to provide a consistent, reliable foundation that simplifies hardware consumption while accelerating deployment at scale.

What’s new today

Today, Red Hat is proud to announce the developer preview of the Red Hat Device Edge for Industrial Controls build, which is specifically tailored for running mission-critical workloads on a highly dependable platform with deterministic performance.

This build takes all the development and learnings of Red Hat Device Edge when tuned and configured for highly deterministic workloads, and delivers a more easily consumed, pre-tuned platform for workloads that demand the highest level of predictable performance at the lowest-possible response times.

Red Hat Device Edge for Industrial Control aims to accelerate both development of next generation “control-style” workloads, as well as ease the adoption with industrial customers, greatly increasing adoption in the market. Partners can benefit from a highly dependable platform pre-built and pre-configured from Red Hat, while customers will have a simplified journey to production deployments for critical workloads.

Industrial community building with Linux

In addition, industrial communities have begun to standardize on Linux, moving away from more traditional proprietary operating systems and heavily embedded systems to seek more resilient and flexible options. Linux platforms, such as RHEL delivered through Red Hat Device Edge, have been meeting the requirements and seeing adoption and deployment in the industrial space, even for industrial control workloads.

The Open Process Automation Forum has been working on next-generation architectures for industrial control systems,  and has included leveraging Linux and containerization platforms as part of their reference architectures. This allows for much greater flexibility to meet the needs of existing systems, while enabling next-generation software stacks that are more modular and have more integration points.

In addition, Margo, an initiative focused on device interoperability and unified deployment of devices and workloads, exists under the Linux Foundation umbrella, and heavily utilizes standards and patterns already in use for modern workload deployments - all powered by a core foundation of Linux.

Finally, large vendors such as Schneider Electric and Codesys both ship and support Linux-based software control runtimes, providing flexibility for their customers, yet trusting in the solid foundation provided by Linux.