Sponsored Keynote: Cloud Native is Edge Native Too! - Dr. Thomas Di Giacomo

Hi everyone and thank you for joining this 'Cloud Native is Edge Native' short discussion. I'm Thomas Giacomo from SUSE, and the one thing I both cannot ignore and cannot talk about today is SUSE and Rancher, the chameleon and the cow, or the so-called chameleon.

Now that it's out of the way, let me share how cloud native technologies, many coming from the CNCF community efforts, are applicable to edge use cases. Edge computing has many applications across industries, and its adoption is growing. Similar to cloud computing, edge computing requires infrastructure, application, and data management, and above all, a smooth integration with on-prem, private data centers, and public clouds. While the as-a-service approach of clouds and the device-based perspective of edge may seem different, they are co-dependent and interconnected for success.

So let's take a concrete example with smart, almost self-driving cars. Bear with me; the forest picture will make sense soon as an edge and cloud visualization. Cars can be seen as endpoints in this edge architecture, represented as leaves in the forest—green because they are smart and clean. Smart cars generate tons of data. Some is processed locally due to latency concerns for driving reaction time. Other data goes back to the branches, meaning on-prem or co-located data centers of automotive manufacturers, where it is stored, analyzed, processed, and filtered. This is a two-way street where leaves get updates from branches. It's also a multiple-way street with exchanges to and from trunks, which are public clouds hosting centralized data and user applications. Last but not least, the ground or underground is where seeds fertilize and support the ecosystem's health—representing users, companies, IT and developer professionals, and tools for managing applications, infrastructure, security, observability, and more. This layer makes the entire forest possible.

Cloud native technologies are very relevant for edge use cases, either directly or with adaptations, helping connect the dots with compute. For instance, Linux and containers can run virtually anywhere. To illustrate, I'll share a few examples of projects and where they fit in the forest, as each layer has different purposes and requirements. The leaves, or edges, can be sensors, gateways, or small data centers—usually more numerous and smaller, requiring optimized, lightweight solutions. Technologies like KubeEdge or K3s enable edge-friendly Kubernetes environments, and Longhorn provides edge-friendly persistent storage back-end. Networking, connectivity, and operating systems with special Linux flavors for edge on specific hardware and chipsets, such as ARM, are needed at this level. Moving to branches, or on-prem data centers, with more resources and traditional setups, Rook with Ceph is a relevant storage backend, and full-fledged Kubernetes distributions can sit here. For trunks, public cloud services like AKS, EKS, GKE, and others are used. All layers must integrate well; if any part is unhealthy or poorly implemented, the entire forest is at risk. On the ground, we need to abstract the ecosystem while managing all living, evolving components. Edge computing networks grow fast and include diverse clouds and devices, but projects provide multi-cloud, multi-cluster, and edge management capabilities to cope.

In closing, remember that edge computing requires integrated infrastructure spanning edge to core to cloud. Cloud native technologies are being adapted for edge computing, and these advances will enable Kubernetes and cloud native technologies to become the unifying platform making edge computing available to the world. I hope you take advantage of your time at KubeCon to learn more about these projects, and remember that cloud native is really edge native too. Thank you.