Terraform at Scale: Lessons from Managing 500+ Resources

Short summary

When a single Terraform state file manages 500+ resources, plans take 8+ minutes and state conflicts become daily. Split infrastructure into 6 domain-focused state files using remote state references, and implement Terragrunt for environment consistency. Enforce CI/CD-only applies with DynamoDB locking—this team reduced plan time to 45 seconds and eliminated conflicts.

• •Split monolithic state into 6 domain-focused files (network, compute, data, DNS, IAM, monitoring) to parallelize plans and reduce blast radius
• •Use Terragrunt and remote state datasources to enforce environment parity across dev/staging/prod, eliminating copy-paste drift
• •Enforce CI/CD-only applies with DynamoDB state locking to eliminate daily state conflicts

Generated with AI, which can make mistakes.