Declarative vs Imperative Why Terraform Works the Way It Does

​

Declarative vs Imperative — the core idea

• Declarative: describe the desired final state of your infrastructure (the “what”). Terraform figures out the sequence of actions required to reach that state.
• Imperative: provide explicit step-by-step commands that perform changes in a specific order (the “how”).

​

High-level examples

• You declare: “I want an EC2 instance with these properties.”
• Terraform computes: which API calls to make, in what order, and what must change to reach that state from the current state.

• You write a script with ordered steps: create VM, SSH to configure it, attach it to the network, run commands.
• The script author must manage ordering, error handling, retries, idempotency, and safe re-runs.

​

Concrete examples

provider "aws" {
  region = "us-east-1"
}

resource "aws_instance" "example" {
  ami           = "ami-0c94855ba95c71c99"
  instance_type = "t2.micro"

  tags = {
    Name = "example"
  }
}

• This configuration describes the desired end state: an EC2 instance with the specified AMI, instance type, and tag.
• terraform plan compares the desired configuration with the saved state and the live infrastructure (after refresh), showing a preview of changes.
• terraform apply executes the API calls needed to create or modify resources to match the configuration.

#!/bin/bash
# create VM
aws ec2 run-instances --image-id ami-0c94855ba95c71c99 --count 1 --instance-type t2.micro --tag-specifications 'ResourceType=instance,Tags=[{Key=Name,Value=example}]'

# configure VM (example)
ssh ec2-user@1.2.3.4 'sudo yum install -y httpd; sudo systemctl enable --now httpd'

# attach to network/security group (example)
aws ec2 modify-instance-attribute --instance-id i-0123456789abcdef0 --groups sg-0123456789abcdef0

• This script explicitly enumerates actions and their order.
• The author must make the workflow idempotent, handle ordering and transient errors, and ensure the script is safe to re-run.

​

Why Terraform’s declarative model helps

• Plan and Preview: terraform plan computes a change plan so you can review intended modifications before applying them.
• Resource Graph: Terraform builds a dependency graph from your configuration and can parallelize independent operations while respecting dependencies.
• Idempotency: Re-applying the same configuration should converge to the same target state; idempotency depends on provider behavior and external changes.
• Drift detection: Terraform refreshes state and compares configuration and saved state to real-world resources to detect drift.
• Fewer accidental mistakes: You express the desired end state rather than scripting every step; this reduces manual ordering mistakes and hidden side effects.

​

Comparison: Declarative vs Imperative

​

What “declarative” does not mean

​

State and caveats

• Remote state backends (e.g., S3, Terraform Cloud) enable locking, collaboration, and safer workflows.
• State is the source of truth for mapping resource IDs and attributes — manual edits or corruption can cause unexpected behavior.
• External changes (out-of-band modifications) require a refresh and may produce drift that Terraform will detect during plan.

​

Practical tips

• Always run terraform plan before terraform apply to review changes.
• Use remote state with locking in team environments to prevent race conditions.
• Avoid imperative provisioners for configuration management when possible; use configuration management tools (Ansible, Chef) or immutable images instead.
• Break large configurations into modules and smaller workspaces to reduce blast radius and improve readability.

​

Summary

• Declarative configuration expresses the desired end state.
• Terraform handles sequencing, dependency resolution, and execution to achieve that state.
• This approach improves maintainability, reduces manual errors, and enables reproducible, auditable infrastructure changes.

​

Links and references

• Terraform documentation — Language and Commands
• Terraform state overview
• Terraform provisioners