Hardware to AWS, EC2 to EKS: Two Years, Two Epic Migrations
summary
Describes two staged migrations: first from on-premise bare metal to AWS EC2, then from EC2 to Amazon EKS (Kubernetes). Covers concrete outcomes and techniques: migrating 6 TB of data, using Terraform and Ansible pipelines to reduce deploy time from ~15 to ~3 minutes, splitting services to improve uptime to 99.9%+, implementing autoscaling/HPA and spot instances for ~30% cost savings, and increasing deployment throughput from 15 to 20 deploys/day; also discusses developer-driven rollout strategies, IP exhaustion issues (~700 addresses), and push vs pull deployment approaches.
AI-written summary · about
chapters (13)
- 0:00Infrastructure migrations
- 1:46Bare metal to AWS
- 8:23EC2 migration challenges
- 10:29AWS cost explosion
- 14:19Why EKS migration
- 18:10Planning EKS migration
- 24:14Testing EKS performance
- 26:01Rolling out EKS
- 29:30EKS migration insights
- 33:22Tuning EKS for Java/Scala
- 37:13EKS migration results
- 42:29Lessons learned
- 43:30Q&A
related talks
43mReplatform to Amazon EKS
23mParadox's EKS Migration
36mMigrating from kOps to EKS with Minimal Developer Involvement
29mFrom EC2 to Kubernetes Takeaways from a Brownfield Migration
33mNavigating the Seas of Transformation: Migrating from EC2 to Kubernetes
48mA road trip into AWS and back to bare metal
09mHow migrating from ECS to EKS supercharged our infrastructure
46mMigrating from VMs to K8s