Scaling Terraform at
summary
ThousandEyes moved from duplicate Terraform repos and 5,100 .tf files to a code-preprocessor called Stacks that keeps Terraform state sizes small while eliminating boilerplate and making drift obvious. Stacks runs before Atlantis, outputs per-layer deployments in parallel (50 layers that used to be 50 minutes become ~1 minute), injects backends/providers/variables, supports cascading variable scope, secret injection with public/private keys and Jinja templating for module sources, and enforces uniform config across 50 AWS accounts and ~20 regions.
AI-written summary · about
chapters (10)
- 0:00Terraform scaling challenges
- 1:23The problem of slow plans
- 3:33Managing drift and complexity
- 6:19Limitations of current tools
- 8:59Introducing Stacks
- 11:58How Stacks works
- 14:05Developer experience gains
- 18:06Advanced Stacks features
- 21:06Reference architecture
- 23:07Summary and open source
related talks
25mTerraform Stacks is here!
39mTerraform Practices to Enable Infrastructure Scaling
41mScaling Infrastructure as Code
30mBuild and manage complex application platforms with Terraform
30mTerraform Practices to Enable Infrastructure Scaling
59mManaging Terraform modules at scale
31mTerraform Practices to Enable Infrastructure Scaling
48mCodifying Infrastructure with Terraform for the Future