Managing Database Containers with Kubernetes Operators and Docker

In this talk, we will discuss how Kubernetes Operators and Docker work together to manage database containers, focusing on PostgreSQL, MySQL, and MongoDB.
We will break down the complexity and show how these tools simplify managing stateful applications.

I'll also introduce Percona Everest, an open-source tool we're developing at Percona. It helps manage different databases on Kubernetes clusters using a graphical user interface. Check it out here: Percona Everest on GitHub https://github.com/percona/everest

1. Managing Database Containers with Kubernetes Operators and Docker

2. Technology Evangelist at Move UK Global Talent Visa CNCF Ambassador Docker Captain Open Source Contributor: Kubernetes Website Kubernetes Shadow Release 1.3 Edith Puclla Edith Puclla @edithpuclla

3. Docker Kubernetes Deploying an Application Default Resources Limitations Kubernetes Operators Components CRD, CR, OLM, Controllers Operator framework, Operator Hub, Capability Models Agenda

4. Containers OPERATING SYSTEM PROCESS Container

5. Kubernetes Deployment automation Scaling based on demand Application portability Self-healing Good option for microservices Active community and wide adoption

6. Kubernetes Terminology Pods Deployments Services

7. Building a Snack Voting App with Docker and Kubernetes

8. Example: Snack Voting Application voting-app result-app KodeKloud : www.youtube.com/watch?v=XuSQU5Grv1g CHIPS 100% CHOCOLATE 0% CHIPS CHOCOLATE

9. voting-app result-app DB worker Containers Valkey

10. Pods 6379 5432 80 80 POD POD POD POD POD voting-app result-app Valkey DB worker

11. Services 6379 5432 80 80 POD POD POD POD POD voting-app result-app Valkey DB worker Service Service Service Service

12. DEPLOYMENT DEPLOYMENT DEPLOYMENT DEPLOYMENT DEPLOYMENT Service Service POD POD POD POD POD voting-app result-app Valkey DB worker Service Service POD voting-app POD result-app POD voting-app POD result-app Deployments

13. YAML

14. The Architecture Before the Operator Show

15. UI CLI API Server Scheduler Controller-Manager etcd Docker kubelet kube-proxy Container 1 Control Plane User Interface Worker node 1 Container 2 Container 3 Container 1 Container 1 Container 2 Pod 1 Pod 2 Pod 3 Docker kubelet kube-proxy Container 1 Worker node 1 Container 2 Container 3 Container 1 Container 1 Container 2 Pod 1 Pod 2 Pod 3 kubectl

16. Kubernetes Operators

17. Stateless Application Scaling: Easy

18. kubectl scale deploy/staticweb --replicas= 4 kubectl scale deploy/staticweb --replicas= 2

19. What about applications that store data ? Stateful

20. “Deploy” a database : easy POD DB

21. I am the persistent volume! Muahahahaha! Muahaha!

22. Kubernetes Application Lifecycle

23. Running a database over time is the hardest

24. Kubernetes Operators In Action! Super…

25. ‹#› Kubernetes Operators: Extend the Kubernetes API

26. UI CLI API Server Scheduler Controller-Manager etcd Docker kubelet kube-proxy Container 1 Control Plane User Interface Worker node 1 Container 2 Container 3 Container 1 Container 1 Container 2 Pod 1 Pod 2 Pod 3 Docker kubelet kube-proxy Container 1 Worker node 1 Container 2 Container 3 Container 1 Container 1 Container 2 Pod 1 Pod 2 Pod 3 kubectl

27. How the API is Extended: Custom Resource Definitions ( CRD ) Custom controllers for specific applications

28. Pods Services Deployments Kubernetes CookBook

29. Pods Services Kubernetes CookBook Deployments Dragon Fruit Salad

30. Pods Services Kubernetes CookBook Deployments Dragon Fruit Salad Dragon Fruit Salad

31. Custom Resource Definitions (CRD)

32. ‹#› Custom Resource Definitions (CRD) my-crontab.yaml

33. ‹#›

34. Custom Controller Observes changes to resources defined in the CRD in the Kubernetes cluster, detects changes, and reacts to level them out. RECONCILE DESIRED STATE CURRENT STATE == Custom Resource Other K8s objects Optionally Modify

35. Kubernetes With/Without Operators bit.ly/3UGB423

36. How do I create the operators? ‹#›

37. ‹#› Capability Level

38. ‹#›

39. More interesting timeline observations 2015 - Kubernetes is for the brave 2017 - AWS EKS Preview release 2018 - Kelsey Hightower: “Kubernetes supports stateful workloads; I don't.” 2019 - Percona Operators for MySQL and MongoDB released 2019-2021 - Databases on Kubernetes are for the brave 2022 - Data on Kubernetes says 70% of companies run stateful on k8s in production 2023 - Kelsey Hightower admits running DBs on k8s is okay

40. Is it still for the brave? No. But there are some implication. It is not for everything, but anyone can try!

41. ‹#› You can use Kubernetes to run production databases effectively Smaller databases = Kubernetes expertise Need automation + scaling Operators

42. ‹#› Production database in Kubernetes is not a good idea yet Huge databases = New to Kubernetes Performance sensitive Manual deployments

43. ‹#› Percona Operators simplify setting up and maintaining robust, enterprise-grade MySQL, PostgreSQL, and MongoDB clusters on Kubernetes Try Percona Operators: Percona Operator for MySQL P ercona Operator for Mongo DB P ercona Operator for PostgreSQL Open Source github.com/percona

44. Percona Everest open source, cloud-native database platform docs.percona.com/everest Open Source github.com/percona/everest

45. Thank you! edith.puclla @percona.com edithpuclla @edithpuclla