DevConf.US 2022

The hybrid cloud is extending all the way to the edge for better data-driven operations and user experiences. Red Hat Edge helps teams to weave together the underlying infrastructure platform. Now, developers should choose from their tech stack the correct set of tools to get the most of this new platform. Frameworks and runtimes like Quarkus, allow existing cloud developers to reuse their Java knowledge to produce native binaries that can run with resource-constrained devices. Edge connectivity offers different challenges, but Event-Driven Architecture (EDA) helps applications to continue running locally and then synchronize data when connectivity is available. This session will introduce you to these tools and how to create a simple set of data pipelines from the edge to the cloud.

Edge is a space that has all of the buzz right now but for some of us that came from customers to Red Hat from retail and energy verticals it is nothing new… EXCEPT the technology has caught up to some of the needs we had 20 years ago as well as we are seeing a convergence of IT and OT with the introduction of millions of low cost IOT devices.

In this session we will talk about how to: Build and deploy a factory for deploying and managing RHEL for Edge (rpm-ostree) devices using OpenShift
Building Ansible Automation Platform using OpenShift Gitops on OpenShift
Managing devices via Ansible Automation Platform
Using the Ansible Automation Platform to manage the deployment of the OS
Lifecycle the application running in podman as a systemd service using Ansible Automation Platform to manage them at scale.
Declarative and imperative systems and how you need both in order do have completely automated gitops deployments

Audience Takeaways and Ecosystem Benefits:
How people are doing large scale deployments of containerized applications at the edge on small form factor devices at low cost of ownership
Understanding of gitops in imperative and declarative use cases

Imperative vs Declarative Brief(excerpt):
In GitOps, we make a distinction between declarative and imperative systems. The commonly accepted distinction goes as follows:
Imperative Automation defines the steps to execute in order to reach the desired solution.
Declarative Automation defines the desired end state of the automation.
In general, declarative automation systems are considered better because they do more for the user - they abstract more of the automation task by forcing users to consider an end state rather than worry about the details of how to reach that end state (or at least tell users and operators why they cannot). Kubernetes classifies itself as a declarative system.

Micro-edge devices are becoming more popular, maybe an IoT application running in a small RaspberryPI zero or similar SBC. Project Flotta adds a way to manage container workloads with a low footprint on those kinds of devices. Also, it integrates with Fido device initialization, manages the Operation system lifecycle, and has an observability layer that edge administrators can use. In this talk, I'll present how Flotta manages a fleet of edge devices and how to run a sample scenario. https://project-flotta.github.io/

conmon, or container monitor, is a process responsible for watching over containers after they've been configured by a container manager like CRI-O or podman, and before they've been started by an OCI runtime like runc or crun. Initially written in C, it mirrors logs, watches for OOM kills and handles container exiting procedures.

However, with the world moving towards the edge, and pods becoming a ubiquitous form of container operation, conmon is due for an update. And what better language to do so with than in Rust!

conmonrs is a rewrite of conmon, in rust! Some design goals of conmonrs include: updating to one agent per pod, keeping a minimal memory footprint to reduce CRI-O's (and thus Kubernetes's) memory overhead per pod, and taking on some of the management of pods for CRI-O and podman.

Join Peter Hunt in this talk to take a tour of where everyone's favorite container agent has been, and hear about where it's headed.

You've heard of large scale open source architectures, but have you ever wanted to take a serious look at real life enterprise edge implementations that scale? This session takes attendees on a tour of multiple use cases for enterprise challenges on the edge with integration, telco, healthcare, manufacturing, and much more. Not only are these architectures interesting, but they are successful real life implementations featuring open source technologies and power many of your own edge experiences.

The attendee departs this session with a working knowledge of how to map general open source technologies to their own edge solutions. Material covered is available freely online and attendees can use these solutions as starting points for aligning to their own solution architectures. Join us for an hour of power as we talk architecture shop!