Structuring Terraform Modules for AWS Serverless

The Problem

Many Terraform projects start well but quickly become unmanageable: a single monolithic state, copy-paste between environments, plans that take 10 minutes.

Recommended Module Architecture

infra/
├── modules/
│   ├── api-gateway/     # API Gateway module
│   ├── lambda/          # Generic Lambda module
│   ├── dynamodb/        # DynamoDB module
│   ├── monitoring/      # CloudWatch + alerts module
│   └── networking/      # VPC / Security Groups module
├── environments/
│   ├── dev/
│   ├── staging/
│   └── prod/
└── terragrunt.hcl       # Root Terragrunt config

Key Principles

1. One module = one responsibility: separate API Gateway, Lambda, DynamoDB
2. Typed variables: use Terraform's complex types
3. Explicit outputs: each module exposes what others need
4. Terragrunt for DRY: avoid duplication between environments

Reusable Lambda Module

A good Lambda module should accept:

• Source code (S3 or local)
• Memory/timeout configuration
• Environment variables
• Additional IAM policies
• Monitoring configuration

Terragrunt: The Game Changer

Terragrunt eliminates duplication between environments. A single terragrunt.hcl per environment referencing modules with specific variables.

Observed Results

• Terraform plans from 10min to 30s
• New dev onboarding: 1 day instead of 1 week
• Deployments reproducible and auditable
• Automated drift detection

Need to structure your IaC? Contact me.