Download to read offline
Uploaded byEdith Puclla
Managing Database Containers with Kubernetes Operators and Docker
The document discusses managing database containers using Kubernetes and the role of Kubernetes operators in streamlining deployment and scaling. It emphasizes the benefits of Kubernetes for microservices and highlights case studies, including a snack voting app. Additionally, it covers the evolution of Kubernetes capabilities and the applicability of running production databases, while promoting Percona operators for managing complex database setups on Kubernetes.
More Related Content
Similar to Managing Database Containers with Kubernetes Operators and Docker
!["[WORKSHOP] K8S for developers", Denis Romanuk](https://cdn.slidesharecdn.com/ss_thumbnails/k8sfordevs-210906100414-thumbnail.jpg?width=640&height=640&fit=bounds)
Managing Database Containers with Kubernetes Operators and Docker
- 1. Managing Database Containers withKubernetes Operators and Docker
- 2. ● Technology Evangelistat ● 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
- 5.
- 7. Kubernetes ● Deployment automation ●Scaling based on demand ● Application portability ● Self-healing ● Good option for microservices ● Active community and wide adoption
- 8.
- 9. Building a SnackVoting App with Docker and Kubernetes
- 10. Example: Snack VotingApplication voting-app result-app KodeKloud:www.youtube.com/watch?v=XuSQU5Grv1g CHIPS 100% CHOCOLATE 0% CHIPS CHOCOLATE
- 11.
- 12.
- 13.
- 14.
- 16.
- 17. The Architecture Beforethe Operator Show
- 18. UI CLI API Server Scheduler Controller-Manager etcd Docker kubelet kube-proxy Container1 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
- 19.
- 20.
- 21. kubectl scale deploy/staticweb--replicas=4 kubectl scale deploy/staticweb --replicas=2
- 22. What about applicationsthat store data?
- 23. “Deploy” a database:easy POD DB
- 24.
- 25.
- 26. Running a databaseover time is the hardest
- 27.
- 28. 28 Kubernetes Operators: Extend theKubernetes API
- 29. UI CLI API Server Scheduler Controller-Manager etcd Docker kubelet kube-proxy Container1 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
- 30. How the APIis Extended: ● Custom Resource Definitions (CRD) ● Custom controllers for specific applications
- 33.
- 34.
- 35.
- 36. 36 Custom Resource Definitions(CRD) my-crontab.yaml
- 37.
- 38. Custom Controller Observes changesto resources defined in the CRD in the Kubernetes cluster, detects changes, and reacts to level them out. RECONCILE DESIRED STATE CURRENT STATE == CustomResource Other K8s objects Optionally Modify
- 39.
- 40. How do Icreate the operators? 40
- 42.
- 43.
- 44.
- 45. More interesting timelineobservations ● 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
- 47. Is it stillfor the brave? No. But there are some implication. It is not for everything, but anyone can try!
- 48. 48 You can use Kubernetesto run production databases effectively Smaller databases = Kubernetes expertise Need automation + scaling Operators
- 49. 49 Production database in Kubernetesis not a good idea yet Huge databases = New to Kubernetes Performance sensitive Manual deployments
- 50. 50 Percona Operators simplifysetting up and maintaining robust, enterprise-grade MySQL, PostgreSQL, and MongoDB clusters on Kubernetes Try Percona Operators: ➔ Percona Operator for MySQL ➔ Percona Operator for MongoDB ➔ Percona Operator for PostgreSQL github.com/percona
- 51. Percona Everest opensource, cloud-native database platform ➔ docs.percona.com/everest github.com/percona/everest
- 53. Thank you! edith.puclla@percona.com edithpuclla@edithpuclla
Editor's Notes
- #47 Keysey knew we needed to work harder on these challenges, and over time it became possible and mature for us to run stateful applications on Kubernetes. —--------------------------------- Essentially, Kubernetes can facilitate database deployment and provide some level of automation and scalability, but it doesn't automatically provide the extensive managed services features that come with solutions like Cloud SQL. Understanding and implementing these additional features in Kubernetes require more sophisticated configuration and management, that is the "biggest challenge."