Natalie Godec and Endy Kasanardjo had a rock-solid microservices platform on AWS—Kubernetes, GitLab, Flux2, Prometheus, Terraform, Kafka, Postgres and even a lightweight LinkerD mesh—complete with fully automated CI/CD and clean data pipelines. Everything worked beautifully, until the day they decided to move it all to GCP.
What should’ve been a smooth “lift and shift” turned into a series of “oops” moments: cluster quirks, data replication fails, Prometheus and fluentd behaving differently, and sneaky DNS pitfalls. Their talk dives into the painful re-engineering, clever workarounds, and final triumphant DNS flip that made the GCP migration stick.
Watch on YouTube
Top comments (0)