Red Hat unveiled a new Kubernetes-based edge device platform that took use case ideas from Lockheed Martin and ABB in stripping down its OpenShift container orchestration platform to better target resource-constrained edge locations

Nick Barcet, senior director of technology strategy at Red Hat, explained that the Device Edge platform is the vendor’s enterprise-focused and supported distribution of its previously unveiled MicroShift project. MicroShift was launched earlier this year as a way to provide a slimmer container orchestrator targeted at limited resource environments.

Barcet noted Device Edge was based on developing a platform for specific use cases brought up by Lockheed Martin and tech vendor ABB.

The Lockheed Martin use case was about providing direct control to drones that had already been deployed but might need to have their mission altered in flight.

“For example, a boat sinks out of the coast of California and suddenly you instead of doing image recognition from the camera to look for fires, suddenly you need to look for people that are at sea because you're gonna need to rescue them,” Barcet said. “And that, by itself, justifies having the ability to be able to switch the workload that is running on the small devices is running inside of the drone and doing real-time image analysis from the camera from one model to the other.”

The ABB use case was about being able to scale data analysis between connected cloud and control edge environments.

MicroShift was initially designed to provide the same Kubernetes interface from OpenShift for “field-deployed edge computing devices.”

“Field-deployed devices are often single-board computers chosen based on performance-to-energy/cost ratio, usually with lower-end memory and CPU options,” Red Hat explained in the initial MicroShift launch. “These devices are centrally imaged by the manufacturer or the end user’s central IT before getting shipped to remote sites such as roof-top cabinets housing 5G antennas, manufacturing plants, etc.”

Barcet said that beyond having a smaller footprint, MicroShift taps into the vendor’s Enterprise Linux (RHEL) platform to tackle container orchestration differently than OpenShift.

“We started to realize that the problem went beyond just the problem of Kubernetes, that you needed to handle both the problem of the OS management and the container management,” Barcet said. “In the process of provisioning the solution to their customer, [Lockheed Martin and ABB] needed a model that was really different from the traditional data center model. And that's where the idea of combining the features that we were putting into MicroShift with the feature that we were putting into RHEL in order to solve for edge functionality into a single product.”

Red Hat earlier this year introduced RHEL 9, which is the latest iteration of its long-running enterprise Linux platform. That update included a technology preview of the FIDO Device Onboarding that automates some of the deployment steps at edge locations and creates a security channel for the configuration and onboarding to a management platform.

The Red Hat Device Edge platform also furthers what has been a growing push to squeeze Kubernetes functionality into resource-constrained environments. This includes efforts around the Rancher Labs’ developed K3s Kubernetes distribution and the KubeEdge open source project that supports edge nodes, applications, devices, and cluster management consistent with the Kubernetes interface.

Barcet did note that MicroShift shared some similarities with these other platforms.

“We didn't start from nothing,” Barcet said. “We looked at what was on the market and we realized that a good way to save on the footprint of Kubernetes was to compile Kubernetes into a single binary. So we use that same model, and from that perspective MicroShift and K3s are very similar.”

What’s different is that MicroShift “is a distribution of Kubernetes and not a fork,” Barcet said. “It's not a separate project from Kubernetes, 100% of the code that is in MicroShift is upstream Kubernetes. There is no difference, nothing that we've removed from the Kubernetes code, nothing that we added. We just recompile it.”

The Device Edge platform is scheduled for developer previews early next year, with general availability and full support expected later in 2023.