If You Can, It Doesn't Mean You Should: Lessons From Terraforming Clouds
summary
Argues that just because Terraform can automate something doesn’t mean it should — packed modules and procedural hacks create fragile infra and surprise failure modes like dependency loops in 2000 resources or misconfigured fleets across 400 databases. Calls out hard cases (VPC peering, cloud Kubernetes, OpenSearch/Cloud Composer) and prescribes grouping deployable units into a single point of deployment, plus reusable patterns and guardrails such as Checkov and org policies to keep things simple and maintainable.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:01Automation and abstraction layers
- 3:04Pitfalls of over-engineering Terraform
- 6:34All-in-one states and security risks
- 8:18Single point of deployment concept
- 11:05Design patterns and guardrails
- 13:33Summary and KISS principle
related talks
30mSturdy Clouds Creating Resilient Software Defined Data Centers
27mStriving for simplicity: Creating software that is simple to use and easy to maintain
20mBuilding a cloud-friendly application
29mThe Future of the Cloud
38mIt's a Trap! How Abstractions Have Failed Us.
57mBuilding Systems that are #neverdone
05mMake it right, from the Start: how clean code and architecture enable cloud
53mChange your tools, change your outcome