The platform is dead, long live the platform
summary
Old ansible + Jenkins + HAProxy + nginx + Icinga gets replaced by container images, Travis CI, ECR, CloudFormation and ECS/Fargate with CloudWatch for logs. The platform moves complexity into Docker builds, supports Java/Python/Go side‑by‑side, autoscaling works, and environments spin up per branch so 100→2000 req/s traffic growth never forces rewrites and pull-request live preview URLs appear automatically. Pain points: shoehorning legacy apps into Docker and CloudFormation quirks (stack rollbacks, numeric listener priorities). Result: self‑service deployments and simplified runtime operations via ECS Fargate + CloudFormation.
AI-written summary · about
chapters (9)
- 0:00Introduction and platform overview
- 1:03Legacy platform stack and its success
- 7:53Why the legacy platform failed
- 9:40Moving to containers and Travis
- 12:51Choosing ECS and CloudFormation
- 15:46Networking and security setup
- 17:54Application deployment workflow
- 20:43Challenges and hurdles faced
- 22:13Q&A session
related talks
28mWhat's Next in Platform Engineering
49mThe road to continuous deployment: a case study
15mThe Art of Platform Engineering
36mPractical Cloud Native: What Works, What Doesn't
44mPlatform Engineering in action
40mGoes Beyond CI to Production Scale PaaS with Docker
25mBreaking the 2 Pizza Paradox with Platform Applications
29mYour platform is not an island: building a platform for your ecosystem