Terraform Best Practices

Coding standards and best practices for writing maintainable, scalable, and reliable Terraform configurations.

Project Structure

Recommended Directory Layout

terraform/
├── environments/
│   ├── dev/
│   │   ├── main.tf
│   │   ├── variables.tf
│   │   ├── outputs.tf
│   │   ├── terraform.tfvars
│   │   └── backend.tf
│   ├── staging/
│   └── production/
├── modules/
│   ├── networking/
│   │   ├── main.tf
│   │   ├── variables.tf
│   │   ├── outputs.tf
│   │   └── README.md
│   ├── compute/
│   └── database/
├── global/
│   ├── iam/
│   └── route53/
└── README.md

File Organization

Standard Files:

• main.tf - Primary resource definitions
• variables.tf - Input variable declarations
• outputs.tf - Output value declarations
• versions.tf - Terraform and provider version constraints
• backend.tf - Backend configuration
• locals.tf - Local value definitions (if many)
• data.tf - Data source definitions (if many)
• terraform.tfvars - Variable values (not committed for secrets)

When to Split Files:

• More than 200 lines in a single file
• Logical grouping of resources (e.g., networking.tf, compute.tf)
• Complex modules with many resource types

Naming Conventions

Resources

Pattern: <resource-type>_<descriptive-name>

# Good
resource "aws_instance" "web_server" {}
resource "aws_s3_bucket" "application_logs" {}
resource "aws_security_group" "database_access" {}

# Avoid
resource "aws_instance" "instance1" {}
resource "aws_s3_bucket" "bucket" {}

Variables

Pattern: snake_case with descriptive names

# Good
variable "vpc_cidr_block" {}
variable "instance_type" {}
variable "environment_name" {}

# Avoid
variable "VPCCIDR" {}
variable "type" {}
variable "env" {}

Modules

Pattern: kebab-case for directories, snake_case for module calls

# Directory: modules/vpc-networking/

module "vpc_networking" {
  source = "./modules/vpc-networking"
}

Tags

Consistent Tagging Strategy:

locals {
  common_tags = {
    Environment = var.environment
    ManagedBy   = "Terraform"
    Project     = var.project_name
    Owner       = var.owner_email
    CostCenter  = var.cost_center
  }
}

resource "aws_instance" "web" {
  # ... other config ...

  tags = merge(local.common_tags, {
    Name = "${var.environment}-web-server"
    Role = "webserver"
  })
}

Variable Management

Variable Declarations

Always Include:

• Type constraints
• Descriptions
• Validation rules (when applicable)
• Default values (for non-sensitive, non-environment-specific values)

variable "instance_type" {
  description = "EC2 instance type for web servers"
  type        = string
  default     = "t3.micro"

  validation {
    condition     = contains(["t3.micro", "t3.small", "t3.medium"], var.instance_type)
    error_message = "Instance type must be t3.micro, t3.small, or t3.medium."
  }
}

variable "vpc_cidr" {
  description = "CIDR block for VPC"
  type        = string

  validation {
    condition     = can(cidrhost(var.vpc_cidr, 0))
    error_message = "VPC CIDR must be a valid IPv4 CIDR block."
  }
}

variable "db_password" {
  description = "Database master password"
  type        = string
  sensitive   = true  # Prevents display in logs
}

Variable Types

Use Specific Types:

# Primitive types
variable "instance_count" {
  type = number
}

variable "enable_monitoring" {
  type = bool
}

# Collection types
variable "availability_zones" {
  type = list(string)
}

variable "tags" {
  type = map(string)
}

# Object types
variable "database_config" {
  type = object({
    engine         = string
    engine_version = string
    instance_class = string
    allocated_storage = number
  })
}

Environment-Specific Variables

Use .tfvars Files:

# environments/dev/terraform.tfvars
environment     = "dev"
instance_type   = "t3.micro"
instance_count  = 1
enable_backup   = false

# environments/production/terraform.tfvars
environment     = "production"
instance_type   = "t3.large"
instance_count  = 3
enable_backup   = true

Module Design

Module Best Practices

Single Responsibility: Each module should have one clear purpose.

# Good: Focused module
module "vpc" {
  source = "./modules/vpc"
  # VPC-specific config
}

# Avoid: Kitchen-sink module
module "infrastructure" {
  source = "./modules/everything"
  # VPC, databases, compute, monitoring, etc.
}

Required vs Optional Variables:

# modules/database/variables.tf

# Required - no default
variable "database_name" {
  description = "Name of the database"
  type        = string
}

# Optional - has sensible default
variable "backup_retention_days" {
  description = "Number of days to retain backups"
  type        = number
  default     = 7
}

Output Everything Useful:

# modules/vpc/outputs.tf

output "vpc_id" {
  description = "ID of the VPC"
  value       = aws_vpc.main.id
}

output "private_subnet_ids" {
  description = "List of private subnet IDs"
  value       = aws_subnet.private[*].id
}

output "public_subnet_ids" {
  description = "List of public subnet IDs"
  value       = aws_subnet.public[*].id
}

Module Documentation

README.md Template:

# VPC Module

Creates a VPC with public and private subnets across multiple availability zones.

## Usage

```hcl
module "vpc" {
  source = "./modules/vpc"

  vpc_cidr             = "10.0.0.0/16"
  availability_zones   = ["us-east-1a", "us-east-1b"]
  environment          = "production"
}

Requirements

Inputs

Outputs

## State Management

### Remote State

**Always Use Remote State for Teams:**

```hcl
terraform {
  backend "s3" {
    bucket         = "company-terraform-state"
    key            = "production/vpc/terraform.tfstate"
    region         = "us-east-1"
    encrypt        = true
    dynamodb_table = "terraform-state-locks"

    # Workspace-specific state
    workspace_key_prefix = "workspaces"
  }
}

State Locking

DynamoDB Table for S3 Backend:

resource "aws_dynamodb_table" "terraform_locks" {
  name         = "terraform-state-locks"
  billing_mode = "PAY_PER_REQUEST"
  hash_key     = "LockID"

  attribute {
    name = "LockID"
    type = "S"
  }

  tags = {
    Name      = "Terraform State Locks"
    ManagedBy = "Terraform"
  }
}

State Isolation

Separate State Files by Environment and Component:

s3://terraform-state/
├── production/
│   ├── vpc/terraform.tfstate
│   ├── database/terraform.tfstate
│   └── compute/terraform.tfstate
├── staging/
│   ├── vpc/terraform.tfstate
│   └── compute/terraform.tfstate
└── dev/
    └── all/terraform.tfstate

Resource Management

Use Data Sources for Existing Resources

# Instead of hardcoding
resource "aws_instance" "web" {
  subnet_id = "subnet-12345"  # Avoid
}

# Use data sources
data "aws_subnet" "private" {
  filter {
    name   = "tag:Name"
    values = ["${var.environment}-private-subnet"]
  }
}

resource "aws_instance" "web" {
  subnet_id = data.aws_subnet.private.id
}

Resource Dependencies

Implicit Dependencies (Preferred):

resource "aws_instance" "web" {
  subnet_id         = aws_subnet.private.id  # Implicit dependency
  security_groups   = [aws_security_group.web.id]
}

Explicit Dependencies (When Needed):

resource "aws_iam_role_policy" "example" {
  # ... config ...

  # Ensure role exists before attaching policy
  depends_on = [aws_iam_role.example]
}

Count vs For_Each

Use for_each for Map-Like Resources:

# Good: for_each with maps
locals {
  subnets = {
    public_a  = { cidr = "10.0.1.0/24", az = "us-east-1a" }
    public_b  = { cidr = "10.0.2.0/24", az = "us-east-1b" }
    private_a = { cidr = "10.0.3.0/24", az = "us-east-1a" }
    private_b = { cidr = "10.0.4.0/24", az = "us-east-1b" }
  }
}

resource "aws_subnet" "main" {
  for_each = local.subnets

  vpc_id            = aws_vpc.main.id
  cidr_block        = each.value.cidr
  availability_zone = each.value.az

  tags = {
    Name = each.key
  }
}

Use count for Simple Conditionals:

resource "aws_cloudwatch_log_group" "app" {
  count = var.enable_logging ? 1 : 0

  name = "/aws/app/logs"
}

Version Constraints

Terraform Version

terraform {
  required_version = ">= 1.0, < 2.0"
}

Provider Versions

terraform {
  required_providers {
    aws = {
      source  = "hashicorp/aws"
      version = "~> 5.0"  # Allow patch updates, lock minor version
    }

    random = {
      source  = "hashicorp/random"
      version = "~> 3.5"
    }
  }
}

Version Constraint Operators:

• = - Exact version
• != - Exclude version
• >, >=, <, <= - Comparison
• ~> - Pessimistic constraint (allow rightmost version component to increment)

State Management Blocks

Terraform 1.1+ introduced declarative blocks for managing state without manual terraform state commands.

Import Block (Terraform 1.5+)

The import block allows config-driven import of existing resources into Terraform state.

Basic Usage:

# Import an existing VPC

  to = aws_vpc.main
  id = "vpc-0123456789abcdef0"
}

resource "aws_vpc" "main" {
  cidr_block = "10.0.0.0/16"

  tags = {
    Name = "main-vpc"
  }
}

Dynamic Import (Terraform 1.6+):

# Import with expressions
variable "vpc_id" {
  type = string
}

  to = aws_vpc.main
  id = var.vpc_id
}

# Import with string interpolation

  to = aws_s3_bucket.logs
  id = "${var.environment}-logs-bucket"
}

Generate Configuration:

# Generate config for imported resources
terraform plan -generate-config-out=generated.tf

Workflow:

1. Add import block with target resource address and ID
2. Run terraform plan to see what will be imported
3. Add or generate the corresponding resource block
4. Run terraform apply to import
5. Remove the import block after successful import

Moved Block (Terraform 1.1+)

The moved block enables refactoring without manual state manipulation.

Rename a Resource:

# Old: aws_instance.web
# New: aws_instance.web_server

moved {
  from = aws_instance.web
  to   = aws_instance.web_server
}

resource "aws_instance" "web_server" {
  ami           = "ami-12345678"
  instance_type = "t3.micro"
}

Move to a Module:

# Move resource into a module
moved {
  from = aws_vpc.main
  to   = module.networking.aws_vpc.main
}

module "networking" {
  source = "./modules/networking"
}

Move from count to for_each:

# Old: aws_instance.web[0], aws_instance.web[1]
# New: aws_instance.web["web-1"], aws_instance.web["web-2"]

moved {
  from = aws_instance.web[0]
  to   = aws_instance.web["web-1"]
}

moved {
  from = aws_instance.web[1]
  to   = aws_instance.web["web-2"]
}

resource "aws_instance" "web" {
  for_each = toset(["web-1", "web-2"])

  ami           = "ami-12345678"
  instance_type = "t3.micro"

  tags = {
    Name = each.key
  }
}

Rename a Module:

moved {
  from = module.old_name
  to   = module.new_name
}

module "new_name" {
  source = "./modules/compute"
}

Best Practices for moved:

• Keep moved blocks until all team members have applied the changes
• Remove moved blocks after state migration is complete across all environments
• Use descriptive commit messages explaining the refactoring

Removed Block (Terraform 1.7+)

The removed block allows declarative removal of resources from Terraform management.

Remove Without Destroying:

# Stop managing resource but keep it in cloud
removed {
  from = aws_instance.legacy_server

  lifecycle {
    destroy = false
  }
}

Remove and Destroy:

# Remove from state and destroy the resource
removed {
  from = aws_s3_bucket.old_logs

  lifecycle {
    destroy = true
  }
}

Remove Module:

# Remove entire module from management
removed {
  from = module.deprecated_service

  lifecycle {
    destroy = false
  }
}

Use Cases:

• Migrating resource ownership to another team/state
• Removing resources that should persist but not be managed
• Cleaning up after manual resource creation
• Deprecating modules without destroying infrastructure

State Block Comparison

Migration from CLI Commands

Old Way (CLI):

# Import
terraform import aws_vpc.main vpc-12345

# Move
terraform state mv aws_instance.web aws_instance.web_server

# Remove
terraform state rm aws_instance.legacy

New Way (Config-Driven):

# All operations are declarative and version-controlled

  to = aws_vpc.main
  id = "vpc-12345"
}

moved {
  from = aws_instance.web
  to   = aws_instance.web_server
}

removed {
  from = aws_instance.legacy
  lifecycle {
    destroy = false
  }
}

Benefits of Config-Driven Approach:

• Changes are code-reviewed and version-controlled
• Operations are repeatable and documented
• Team collaboration without state file conflicts
• Rollback capability through git history

Code Quality

Use Locals for Computed Values

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

  common_tags = {
    Environment = var.environment
    ManagedBy   = "Terraform"
  }

  # Computed values
  is_production = var.environment == "production"
  instance_type = local.is_production ? "t3.large" : "t3.micro"
}

Dynamic Blocks

Use Sparingly and Only When Necessary:

resource "aws_security_group" "example" {
  name = "example"

  dynamic "ingress" {
    for_each = var.ingress_rules

    content {
      from_port   = ingress.value.from_port
      to_port     = ingress.value.to_port
      protocol    = ingress.value.protocol
      cidr_blocks = ingress.value.cidr_blocks
    }
  }
}

Conditional Resources

# Use count for conditional creation
resource "aws_kms_key" "encryption" {
  count = var.enable_encryption ? 1 : 0

  description = "Encryption key"
}

# Reference with [0] and handle with try()
resource "aws_s3_bucket" "example" {
  # ...

  kms_master_key_id = try(aws_kms_key.encry