Mastering Module Composition in Terraform

Module composition exists to solve the complexity of managing infrastructure as code. It allows you to take multiple composable building-block modules and assemble them together to produce a larger system. This flat style of module usage simplifies the management of dependencies and improves the overall flexibility of your Terraform configurations.

In practice, you start with a root module and create a flat set of resources, using Terraform's expression syntax to define relationships between them. When you introduce module blocks, your configuration becomes hierarchical. However, it's best to keep the module tree flat, with only one level of child modules. For example, you might define a network module and a Consul cluster module like this:

1module "network" {
2  source = "./modules/aws-network"
3  base_cidr_block = "10.0.0.0/8"
4}
5module "consul_cluster" {
6  source = "./modules/aws-consul-cluster"
7  vpc_id = module.network.vpc_id
8  subnet_ids = module.network.subnet_ids
9}

In production, remember that implicit assumptions and guarantees define what data a module expects and what it produces. This means you need to be clear about your input variables, like ami for disk images or recordsets for DNS records. Keep an eye on potential pitfalls, such as tightly coupling your modules, which can hinder future refactoring. The official docs don't call out specific anti-patterns here. Use your judgment based on your scale and requirements.