Terraform Examples for MLOps Training

This directory contains practical examples demonstrating various Terraform concepts from basic to intermediate levels.

Available Examples

Example	Description	Difficulty
data-sources	Query existing AWS resources and data	Beginner
for-each	Create multiple resources with for_each and count	Intermediate
locals	Use local values for reusable expressions	Intermediate
outputs	Define and format outputs for various use cases	Beginner
modules	Create reusable infrastructure components	Intermediate
conditionals	Environment-specific configurations with logic	Intermediate
lifecycle	Safe resource updates and protection	Intermediate
remote-state	S3 backend for teams	Intermediate

Recommended order: data-sources → outputs → locals → for-each → modules → conditionals → lifecycle → remote-state

---

How to Use These Examples

Prerequisites

1. Terraform 1.14+ installed
2. AWS account with credentials configured
3. Basic understanding of Terraform from the main README

Running an Example

# Navigate to the example directory
cd module-01/terraform/examples/data-sources

# Initialize Terraform
terraform init

# Review the plan
terraform plan

# Apply (only if you want to create resources)
terraform apply

---

Example Descriptions

1. Data Sources (data-sources/)

Learn how to:

• Query AWS account and region information
• Find existing VPCs, subnets, and security groups
• Get the latest AMI IDs dynamically
• Query S3 buckets and IAM roles
• Filter resources by tags and attributes

Key Concepts:

• data blocks
• Filtering with filter blocks
• Referencing data sources in resources
• Using try() for safe access

---

2. For Each & Count (for-each/)

Learn how to:

• Create multiple resources from a list using count
• Create resources from maps using for_each
• Use dynamic blocks with for_each
• Handle nested configurations
• Choose between count and for_each

Key Concepts:

• count meta-argument
• for_each meta-argument
• Dynamic blocks
• Resource references with indices and keys
• Flattening nested structures

---

3. Locals (locals/)

Learn how to:

• Define reusable values within a module
• Create naming conventions
• Build complex tag combinations
• Use conditionals in locals
• Transform and filter data with locals

Key Concepts:

• locals blocks
• DRY (Don't Repeat Yourself) principle
• String manipulation and regex
• Merge and transform functions
• Environment-specific configurations

---

4. Outputs (outputs/)

Learn how to:

• Define basic outputs
• Create complex nested outputs
• Format outputs for human consumption
• Mark sensitive outputs
• Generate JSON and file outputs

Key Concepts:

• output blocks
• Descriptions and formatting
• Sensitive outputs
• Template rendering
• Cross-module outputs

---

5. Modules (modules/)

Learn how to:

• Organize code into reusable modules
• Define module inputs and outputs
• Compose multiple modules
• Use modules for ML infrastructure patterns

Key Concepts:

• Module structure and files
• Module inputs (variables)
• Module outputs
• Module composition for ML workloads

ML Use Cases:

• Reusable ML model storage module
• Training data module with intelligent tiering
• Artifacts module for pipeline outputs

---

6. Conditionals (conditionals/)

Learn how to:

• Use ternary operators for value selection
• Conditionally create resources with count
• Filter resources with for_each conditionals
• Use dynamic blocks with conditionals
• Validate inputs with conditionals

Key Concepts:

• Ternary operator: condition ? true_value : false_value
• Conditional count: condition ? 1 : 0
• Filtered for_each: for_each = {for k, v in map : k => v if v.condition}
• Dynamic blocks with conditional iteration

ML Use Cases:

• Environment-specific instance types
• Production-only lifecycle rules
• Conditional monitoring and alerts
• Account-based compliance levels

---

7. Lifecycle (lifecycle/)

Learn how to:

• Use create_before_destroy for zero-downtime updates
• Protect critical resources with prevent_destroy
• Ignore dynamic changes with ignore_changes
• Force replacement with replace_triggered_by

Key Concepts:

• create_before_destroy - Safe updates
• prevent_destroy - Deletion protection
• ignore_changes - Skip drift detection
• replace_triggered_by - Conditional replacement

ML Use Cases:

• Zero-downtime model deployment
• Production model storage protection
• Ignoring ML pipeline tags
• Replacing resources on model version change

---

8. Remote State (remote-state/)

Learn how to:

• Configure S3 as a remote state backend
• Use modern S3 lock file (no DynamoDB needed)
• Enable state encryption and versioning
• Work with both AWS and LocalStack

Key Concepts:

• S3 remote state backend
• use_lockfile option
• State security and recovery
• LocalStack for local development

---

Common Patterns

Pattern 1: Dynamic Resource Naming

locals {
  name_prefix = "${var.project_name}-${var.environment}"
}

resource "aws_s3_bucket" "example" {
  bucket = "${local.name_prefix}-models"
}

Pattern 2: Conditional Resource Creation

resource "aws_cloudwatch_log_group" "example" {
  count = var.enable_logging ? 1 : 0
  name  = "/aws/ml/example"
}

Pattern 3: Iterating Over a Map

resource "aws_s3_bucket" "buckets" {
  for_each = {
    "dev" = "ml-dev-data"
    "prod" = "ml-prod-data"
  }
  bucket = each.value
}

Pattern 4: Combining Tags

locals {
  common_tags = {
    Project     = var.project_name
    Environment = var.environment
  }

  specific_tags = {
    Purpose = "ml-storage"
  }

  all_tags = merge(local.common_tags, local.specific_tags)
}

resource "aws_s3_bucket" "example" {
  bucket = "example-bucket"
  tags   = local.all_tags
}

---

Practice Exercises

Exercise 1: Create Multi-Environment S3 Buckets

Using the for-each example as reference:

• Create S3 buckets for dev, staging, and prod
• Apply different lifecycle rules for each environment
• Output all bucket names and ARNs

Exercise 2: Build a Naming Convention

Using the locals example as reference:

• Create a locals block with naming conventions
• Apply the naming convention to multiple resources
• Ensure all resources have consistent tags

Exercise 3: Query and Use Existing Resources

Using the data-sources example as reference:

• Find an existing VPC in your account
• Query the latest Amazon Linux AMI
• Create an EC2 instance in the VPC with the AMI

Exercise 4: Build a Reusable Module

Using the modules example as reference:

• Create a simple module for ML model storage
• Define input variables for customization
• Output the bucket name and ARN
• Use the module in your root configuration

Exercise 5: Environment-Specific Configuration

Using the conditionals example as reference:

• Create different instance types per environment
• Add production-only monitoring resources
• Implement conditional tagging

Exercise 6: Safe Resource Updates

Using the lifecycle example as reference:

• Create a resource with create_before_destroy
• Add prevent_destroy to critical storage
• Use ignore_changes for dynamic tags

Exercise 7: Configure Remote State

Using the remote-state example as reference:

• Set up an S3 bucket for state storage
• Configure the S3 backend with encryption
• Enable state locking

---

Tips for Success

1. Start with terraform plan - Always review before applying
2. Use descriptions - Document your outputs and variables
3. Check the state - Use terraform show to inspect results
4. Clean up - Run terraform destroy when done practicing
5. Experiment - Modify the examples and see what happens

---

Next Steps

After mastering these examples:

1. Complete the main basics exercises
2. Work through the comprehensive exercises
3. Build your own MLOps infrastructure from scratch
4. Explore Terraform Cloud/Enterprise for team collaboration

---

Learning Path

Recommended order for learning:

1. Beginner → data-sources, outputs
2. Intermediate → locals, for-each
3. Advanced → modules, conditionals, lifecycle
4. Production → remote-state

---

Need Help? Refer to the main Terraform Basics guide or main README