Terraspace vs Terragrunt

Terraspace and Terragrunt are quite different beasts. Terragrunt describes itself as a thin wrapper for Terraform. Terraspace is a framework. Their workflow, structure, and designs are completely different.

You may also be interested in this blog post: Terraspace vs Terragrunt vs Terraspace


With terragrunt, you cd into the folder with the terraform source code and run the apply command:

cd /path/to/module && terragrunt apply

With terraspace, you stay at the root of the project and run the up command:

terraspace up STACK

Structural Differences

Terraspace defines a clear structure. You know where your reusable modules and business-specific code should go. Examples:


Terragrunt leaves the structure more open. Though, Terragrunt does provide a terragrunt-infrastructure-live-example repo for guidance. You still and must carefully plan the directory structure ahead of time. Terragrunt’s processing logic will rely on this structure. The example structure looks something like this:


With Terragrunt, different environments and regions are handled by walking up the directory tree, looking for terragrunt.hcl files, and processing the logic in the custom Terragrunt syntax.

With Terraspace, normal tfvars and layering is used. You can use the same infrastructure code to deploy to different environments, regions, accounts, projects, etc.

Native Terraform Constructs

Terragrunt uses the native Terraform HCL language for the main tf files, but it also adds a custom terragrunt HCL language in its terragrunt.hcl. For example, to reuse modules, it’s a custom terragrunt syntax. Example:

terraform {
  source = "git::git@github.com/org/demo"

The Terraform module keyword accomplishes the same result, except natively.

module demo {
  source = "github.com/org/demo"

Terraspace uses the native module keyword and Terraform language constructs generally.

Tfvars Variables

Terragrunt has a custom terragrunt.hcl syntax for input variables to modules. For example:

terraform {
  source = "git::git@github.com/foo/modules.git//path/to/module"

input {
  var1 = "val1"
  var2 = "val2"

Terraspace uses standard Terraform module syntax to set input variables. Something like this:

module demo {
  source = "github.com/org/demo"
  var1   = var.var1
  var2   = var.var2

You then create .tfvars files that get process with layering:


The files are regular terraform tfvar files:

var1 = "val1"
var2 = "val2"


To achieve DRY, Terragrunt uses helpers like path_relative_to_include and find_in_parent_folders in the terragrunt.hcl files. It then walks up the directory tree to include other terragrunt.hcl files from parent folders. It will then use things like generate blocks to generate files like backend.tf back down in the original leaf folder of the directory tree.

Terraspace achieves DRY code by building files from your app/modules, app/stacks and config/terraform folder. It processes things in a clear single direction.

Statefile Management

In Terragrunt, how you organize your folder structure generally decides the state file location. Example:


You then use path_relative_to_include and find_in_parent_folders methods in terragrunt.hcl files up the directory tree to define where the state path lives.

In Terraspace, the statefile is centrally defined in config/terraform/backend.tf and processed with variable substitution. For example, this expression in config/terraform/backend.tf:




Deploying Multiple Stacks

Both Terraspace and Terragrunt offers a way to deploy multiple parts of infrastructure with a single command.

With Terragrunt, you cd into the top-level folder with all the subfolders you want to deploy and run:

cd top-level-folder
terragrunt apply-all

What you choose to deploy is determined by the folder structure. If you want to deploy all modules, you go to the “root” of your hierarchal project folder structure.

With Terraspace, to deploy multiple infrastructure stacks, you can run:

terraspace all up

You can also selectively choose which infrastructure stacks to deploy. Simply provide the stack names. Here’s an example targetting the b2 and b3 stacks.

terraspace all up b2 b3

The subgraph stacks, represented by green nodes, will be deployed. For more info see the docs: Deploy Multiple Stacks and Subgraphs


Terragrunt doesn’t have generators.

Terraspace has generators so you can build things quickly.

terraspace new project infra -p aws --examples
cd infra
terraspace new stack demo
terraspace new module bucket


Terragrunt uses a testing framework called terratest. This allows you test real resources.

Terraspace uses rspec-terraspace as the default test framework plugin. Rspec is a mature testing framework with expressive matchers, test setup/destroy blocks etc. It also allows you to test real resources. Terraspace also introduces the concept of a Test Harness to allow you to customize and control the test setup.


There are plenty of more differences between Terraspace and Terragrunt. We cover a good amount of them here. Overall, these two are quite different.