DevEx | there’s no place like k3s
1 like298 views
The document discusses the implementation of cloud-native tooling and the development journey for developers using solutions like k3s and k3d. It emphasizes the benefits of treating a local machine as a cloud environment for scalable application development, including features like hot-reloading and local container registries. Additionally, it outlines various Kubernetes distributions, their advantages, and the importance of integrating development practices such as GitOps and monitoring within the cloud-native framework.
![Self-Service Cloud Resources
cpu, ram, [ disk, network ]](https://image.slidesharecdn.com/devextheresnoplacelikek3s-220217062536/85/DevEx-there-s-no-place-like-k3s-10-320.jpg)
![Self-Service Kubernetes Resources
cpu, ram, [ disk, network ]](https://image.slidesharecdn.com/devextheresnoplacelikek3s-220217062536/85/DevEx-there-s-no-place-like-k3s-11-320.jpg)
DevEx | there’s no place like k3s
- 1. DevEx | N🚫 place like 🏠 k3s k8s.local with cloud-native tooling Haggai Philip Zagury DevOps GL & TL
- 2. DevOps GL & TL Haggai Philip Zagury My Solution driven approach is based on hands-on and deep understanding of Operating Systems, applications stacks software languages and frameworks, Networking, Cloud and Cloud Native solutions. Open thinking and open techniques ideology - driven by Open Source technologies
- 3. Tend to take our roadmap / work pretty seriously (i hope 😉)
- 4. DevEx | N🚫 place like 🏠 k3s k8s.local with cloud-native tooling
- 5. DevEx A Cloud-Native Another touchpoint In our Development Journey
- 6. What does that mean for developers ? What is Cloud-Native thinking?
- 7. Developer Journey ● Define the workload “gestimation” ● Define the sw language
- 8. Developer Journey ● Define the workload ● Define the scale ● Define the architecture ● Define the sw language ● The cloud will take care of the rest … ● We will declaratively tell it how
- 9. Developer Self-Service Cloud Native | like ordering on amazon …
- 10. Self-Service Cloud Resources cpu, ram, [ disk, network ]
- 11. Self-Service Kubernetes Resources cpu, ram, [ disk, network ]
- 12. Use the Best Arch to get the job done What is it ?
- 13. 🏠 ? 127.0.0.1 ? A Cloud-native desktop :)
- 14. PUT IMAGE HERE AND CROP AS CIRCLE 🏠 Local IaaS A smart Developer ;) ● Develop Operations is by design ● Treat your laptop as the 1st cloud ○ Run your code inside a container ○ Run your container inside a pod ○ Portable & Scalable application development ● Sounds 2 good 2B true …
- 15. PUT IMAGE HERE AND CROP AS CIRCLE 127.0.0.1 | Local IaaS The laptop it A Cloud, kubernetes is the Generic API - K3D is a cloud provider ! … - ContainerD is the “image provider” - Kubectl / Kubernetes API / SDK is your cloud native interface ….
- 16. Kubernetes built for dev’s K3d vs k3s vs Kind vs Microk8s vs Minikube ● They all have their advantages / disadvantages ● They all provide an “all-in-one” kubernetes distro
- 17. K3S ● Lightweight certified Kubernetes distro ● 40MB binary, 512MB memory consumption ● Single process integrated Kubernetes master, Kubelet, and containerd ● SQLite in addition to etcd ● Simultaneously released for x86_64, ARM64, and ARMv7
- 18. K3D ● hot-reloading of code ● Full cluster lifecycle ○ create nodes (servers|agents) ○ single / multi-server clusters ● You can test scaling up / down (HPA / Other) ● Integrated LoadBlancer Container Development booster
- 19. K3D ● hot-reloading of code ● full cluster lifecycle ○ create nodes (servers|agents) ○ single / multi-server clusters ● You can test scaling up / down (HPA / Other) Development booster
- 20. K3D ● hot-reloading of code ● full cluster lifecycle ○ create nodes (servers|agents) ○ single / multi-server clusters ● You can test scaling up / down (HPA / Other) A local Multi-cluster environment
- 21. K3D { LoadBalancer Container } ● hot-reloading of code ● full cluster lifecycle ○ create nodes (servers|agents) ○ single / multi-server clusters ● You can test scaling up / down (HPA / Other) ● Integrated LoadBlancer Container Like A managed service A local Multi-cluster environment
- 22. K3D { Container Registry } ● hot-reloading of code ● full cluster lifecycle ○ create nodes (servers|agents) ○ single / multi-server clusters ● You can test scaling up / down (HPA / Other) ● Integrated LoadBlancer Container ● Run a local container registry Like A managed service A local Multi-cluster environment Local registry (mirror)
- 23. K3D { Container Registry } ● hot-reloading of code ● full cluster lifecycle ○ create nodes (servers|agents) ○ single / multi-server clusters ● You can test scaling up / down (HPA / Other) ● Integrated LoadBlancer Container ● Run a local container registry mirrors: docker.io: endpoint: - "http://mycustomreg.com:5000" configs: "mycustomreg:5000": auth: username: xxxxxx # this is the registry username password: xxxxxx # this is the registry password Local registry(mirror)
- 25. See it in action ! https://grafana.infra.tikalk.dev/explore Tilt Loki & Grafana ArgoCD FLUX Operators Controllers OpenFaas Keda Thanos + Prometheus
- 26. Resources
- 27. PUT IMAGE HERE AND CROP AS CIRCLE 127.0.0.1 | Logging workloads tail -f “cluster-logs” - Log standard kubernetes workload definition - Automatically categorize workload based on our labels P.S you can run both locally log like in prod
- 28. PUT IMAGE HERE AND CROP AS CIRCLE 127.0.0.1 | Monitoring workloads Modern monitoring - Metrics, metrics, metrics - standard kubernetes workload definition - Automatically analyze workload based on our labels P.S you can run both locally monitor like in prod
- 29. PUT IMAGE HERE AND CROP AS CIRCLE 127.0.0.1 | Scaling workloads Event Driven (Reactive) Architecture - Utilize standard kubernetes workload definition P.S you can run both locally - App - Queue BTW: & this is also OLD NEWS ;) …
- 30. PUT IMAGE HERE AND CROP AS CIRCLE 127.0.0.1 | GitOps git -repo - app-sync - Changes are scm based - Automatically categorize workload based on our labels P.S you can run both locally deploy like in prod
- 31. Reduce Frustration | Micro Feedback loops ++ ● Helper-tools / Utilities ● Make (my personal favorite) ● Kubectx / kubens ● Common functions ○ Set cloud provider project / profile ○ Get a secret from credentials store ● IDE -plugins ● IDE - the obvious ○ Auto lint / build / deploy
- 32. Shorter ! Feedback Loops DOD - Definition of Done - How do I know it’s done ? - How does the team know it’s done ? - How does management know it’s done ? - How does NOC know it’s functional ?
- 33. Adopt Capabilities & Best Practices DRY principles - Challenge the controller pattern - Custom types - standard api - Cloud Native workflows - Opinionated workflows
- 34. Operator Pattern - Frameworks - Reusable Architecture styles - GitOps - FaaS - ArgoCD - … - Data pipelines - Streaming
- 35. Crossplane - A native way of provisioning cloud resources - Think of a helm chart which can create an managed database (e.g. RDS) from snapshot …
- 36. PUT IMAGE HERE AND CROP AS CIRCLE 🏠 Local IaaS A smart Developer ;) ● Develop Operations is by design ● Treat your laptop as the 1st cloud ○ Run your code inside a container ○ Run your container inside a pod ○ Portable & Scalable application development ● Sounds 2 good 2B true …