Learnings From Providing A Platform API With Kubernetes And Crossplane
summary
Deutsche Bahn builds a unified platform API using Kubernetes with Crossplane as the control plane, running a couple thousand managed resources and shaping Crossplane pre-1.0. Implementation uses Argo CD GitOps, vcluster/Cluster API for virtual control planes, and custom Crossplane controllers and providers (gitlab, argocd, styra, grafana) that were open sourced. Pain points: CRD sprawl with 700+ AWS APIs, 600+ Azure APIs, slow infra provisioning (minutes to tens of minutes), API rate limits and controller contention, and heavy YAML/composition maintenance.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 0:55Who are we
- 2:05What is Crossplane
- 3:51Kubernetes Example
- 8:12Top 9 Challenges
- 9:05Deployment Model
- 10:06Bootstrapping
- 13:59Compositions
- 16:21Kubernetes
- 18:04Infrastructure
- 19:54Leveraging Kubernetes
- 22:25Crd Maturity
- 25:35GitOps
- 28:23Documentation
- 29:34Contributions
- 30:46Outro
related talks
35mDriving the Platform at Deutsche Bahn with Crossplanes Provider-Argocd
31mThe Crossplane Experience
35mBuilding and Enabling APIs with Crossplane 1
37mPlatformOps with Crossplane
30mGetting to Know Crossplane
05mWhat Can Crossplane Actually Do? A Real World Fiel...
36mCrossplane Intro And Deep Dive - The Cloud Native Control Plane Framework
30mExtending Kubernetes Scheduler for Multi-Cluster and Multi-Cloud Workloads