Kubernetes IoT Edge Working Group, 28 Oct 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keynote: Containers Where YOU Want Them: All the Way to the Edge - Nick Barcet, Red Hat

Description

Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon Europe 2023 in Amsterdam, The Netherlands from April 17-21. Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.

Keynote: Containers Where YOU Want Them: All the Way to the Edge - Nick Barcet, Red Hat

From industrials to defense, healthcare to space, or even drones to cameras, containers can now get even closer to their data sources. However, our customers and partner ecosystem have made it clear that they want to manage applications on small devices exactly as they do in the cloud. Join us to get an early look at the newest, smallest edge solution from Red Hat built in collaboration with the community, that provides the flexibility to deploy containers in a pocket-sized footprint with native integration into an open hybrid cloud using end-to-end Kubernetes. This latest addition to an already broad portfolio, provides consistency all the way to edge devices and gateways so that the business can deploy and manage applications anywhere - all built using familiar tools and all made possible by the open source community.

A

Okay, so thank you very much for having me I'm Nick, barsat, I work for red hat in the city organization and for the past four years, I've been working on edge and I also happen to have been working in open source for the past 20 years. I think something like that.

A

One thing I've seen is that we as a community have the power to transform Industries. A good example is what happened when we worked on containerization and virtualization uh to the Telco industry in the past five years. There is not one Telco that hasn't been impacted by the transformation to vnf and now CNF, most of the 5G deployments that you see today are running on containers, and this is something that we keep on doing. Take a look at what we are doing with multi-tenancy with data science with small form factor kubernetes.

A

uh All this is directly having an impact on all the others, industries that we haven't directly touched and Edge is really a part of it. But one thing I've observed is that it always happened the same way.

A

We first identify a few users that have a real problem to solve people that want to go onto the market with something and here I'm, taking the example of two customers, I've been working with Lockheed Martin and the ABB these two large companies, where the people that made us decide that we should be working on a small form factor of kubernetes.

A

One of them is doing data science, the other one is doing data collection, but they are pretty good example of what uh the deployment of kubernetes at the edge would be and what we did was engage them on working with us on a community project called microshift microshift is a small fun Factor kubernetes doing taking the vanilla Upstream code, compiling it into a single binary, hiding adding a few things here and there.

A

But one of the things we realized when we started working with them is that kubernetes was not everything that these customers wanted and really what was needed was an ability for those devices that we were going to be deploying to behave the same way as the onboard computers or as the uh industrial computers that these guys were already deploying it's a behavior of a device, not the behavior of a server.

A

So we decided that we needed to add functionality into the OS as well, and once we validated that what we've done together worked or flies in this example. Well, that's when we knew that it was time to go and build a product out of it. In this example, right here, Lockheed Martin is using a drone to survey vast areaites. So basically, you can imagine a drone taking off for a multiple hour mission and go and Survey California, for example, to look out for fire starting.

A

But what happens when these drone is flying is that they need to do the detection in real time, so they can't allow two 4K cameras to be sending their feedback to a station in real time. So they need to have handling of the process on the Drone in real time.

A

But the thing is, since the Drone is flying for a long time, condition can change while the Drone is flying, and this use case where you need to switch to workloads. Basically, the AI model that you're using while the Drone is in Flight, is what Justified for us the need to use kubernetes on this drone, because you can run containers in multiple ways, but kubernetes is good at orchestrating.

A

So when you need to be orchestrating, that's when you need kubernetes- and in this case this drone is capable of switching its Mission. Imagine that a boat would sink of the coast of California.

A

While the troll is flying, you can remotely send an order to the Drone to switch to oh, let's look out for people that are droning at Sea, so that we can send rescue for them and now that this drone has flown for the first time last week, it's time for a company such as red hats, to a launch a product like the one we are announcing this morning right at device Edge and what we're doing with red device Edge is taking Upstream kubernetes, combine it with Linux an immutable Linux and delivering something that can run in minimal configurations on arm and x86 without any issues, and this allows us to complete the wide panel of topologies that we were able uh to meet so from the smallest to the largest environment at the edge.

A

You can have your containers everywhere, you want them.

A

So if you want to know more about uh microshift and this project, there are a couple sessions this afternoon at uh openshift. Commons one is a demonstration by Lockheed Martin of the AI they're running on a very small device and another one is a presentation by two of the engineers that have been working on microshift, that is happening. uh Frank, darsky and John cope are two of the main contributors to this project.

A

Thank you very much, and if you want to come and contribute with us here is where you can do that.

A