The Reality of Porting Kubernetes Apps Across Cloud Providers
summary
A demo EKS→GKE migration of demo app moves three microservices and exposes concrete breakages: container images in ECR require credential/workload-identity changes or pushing to GCR, S3 objects need CSI or S3-compatibility on GCS, RDS requires Cloud SQL feature checks or running Postgres in-cluster, and HashiCorp Vault vs GCP Secrets Manager forces secret-model changes. The team wires Dapr sidecar for pub/sub, uses Crossplane composite resources to keep YAML stable, and finds portability works but costs and operational burden rise.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:24Kubernetes portability focus
- 3:53Stakeholders and architecture
- 6:46Migrating container images
- 10:21Database migration challenges
- 14:18Debugging storage APIs
- 18:30Container Storage Interface
- 21:18Message queue migration and Dapr
- 28:42Infrastructure as code with Crossplane
- 33:25Monitoring and validation
- 35:53Key takeaways
related talks
33mKubernetes-Driven Multi-Cloud Deployments: Maximizing Scalability and Flexibility
37mBalancing Power and Pain: Moving a Startup From a PaaS to Kubernetes
41mAn Adventure in Multi-Cloud with Kubernetes
05mMoving to Kubernetes, or not?
35mKubernetes journey
27mKubernetes Everywhere: Lessons Learned From Going Multi-Cloud
32mThe Cloud Provider Lift and Shift: Staying Mobile with Node.js and Kubernetes
35mThe Oregon Trail to Kubernetes [I]