Knowing how to build REST API with latest tech is cool. But you know what is even cooler? - Being able to deploy it to the cloud! I'll walk you through the process of building simple, server-less REST API using GO, AWS Lambda, API Gateway, DynamoDB and then deploy it all to AWS cloud with Terraform. If you want to follow along, you'll need an AWS account with user that has programmatic access enabled, and . Installing these dependencies is relatively easy and there's plenty of online resources to guide you through. I would also recommend looking into to manage multiple versions of terraform locally. GO Terraform tfenv For this project I'm using and . Assuming all dependencies are installed, I can start crafting. go v1.17.1 terraform v1.0.11 Business specific logic will be kept in and infrastructure in (Infrastructure As Code) directory. api iac serverless-api/ |-api/ |-lambdas/ |-internal/ |-iac/ |-prerequisites/ |-modules/ |-api/ Since I'll be using to develop api, I'll be placing my lambdas in and all shared/reusable packages in directory. In above diagram there's directory and this is what I'll be working on first. go /lambdas /internal iac/prerequisites Terraform has the ability to store its state remotely in a variety of back-ends and since deployment will happen on AWS, I'll use S3 for that. This will allow me to easily manage lifecycle of the application as well as allow other participants to share and update same state. That way everything  and everyone will be in sync. Prerequisites is kept separate as it has to be deployed first. This is because our main applications configuration block cannot contain interpolated values and is initialised prior to Terraform parsing the variables. Let's start from the main config file: terraform > backend provider "aws" { } resource "aws_dynamodb_table" "terraform_statelock" { attribute { }
} resource "aws_s3_bucket" "remote_state" { versioning { } }
} # iac/prerequisites/main.tf # ---------------------------------------------------- # configure aws provider # ---------------------------------------------------- region = var.region # ---------------------------------------------------- # create dynamo db for deployments locking # ---------------------------------------------------- name = local.ddb_name read_capacity = 20 write_capacity = 20 hash_key = "LockID" name = "LockID" type = "S" # ---------------------------------------------------- # create S3 bucket that will be used for the backend # ---------------------------------------------------- bucket = local.state_bucket_name force_destroy = local.destroy_bucket acl = "authenticated-read" enabled = true tags = { Env = local.env Here I'm creating S3 bucket to hold shared state and DynamoDB table for locking deployments. Locking is needed for cases when two or more people are trying to perform deployment at almost exact same time and this is where things can get out of sync. Important thing to mention is DynamoDB . For locking to work, it has to be explicitly named and it's case sensitive. hash_key LockID Quick note before moving on. File names do not matter. Terraform will read all configuration files in top-level directory. Nested directories are treated as completely separate modules and will not be automatically included in the configuration. You don't even have to split your logic into variables.tf, locals.tf, main.tf like I do. It could all be placed within a single file. It is simply my personal preference to organise configuration by splitting logical units into separate files. Next , and : variables.tf locals.tf versions.tf } } } # iac/prerequisites/variables.tf variable "region" { default = "eu-central-1" description = "region where all the resources will be deployed" variable "prefix" { default = "project-123" description = "organization or service name, has to be unique" variable "ddb_statelock_table" { default = "tf-statelock" description = "name of dynamo db table for terraform state locking" is the way to define program specific variables, add validations and set defaults. All these values can be conveniently overridden by exporting same variable name and adding a prefix to it. Say if I'd like to use different region in CI/CD pipeline, I'd and this value would take priority. variables.tf TF_VAR_ export TF_VAR_region=us-east-1 locals { } # iac/prerequisites/locals.tf env = terraform.workspace == "default" ? "dev" : terraform.workspace state_bucket_name = "${var.prefix}-remote-state" destroy_bucket = contains([ "prod" , "staging" ], local.env) is used to dynamically compute variables and avoid duplicate logic in rest of configuration. locals.tf terraform {
  required_providers { }
  } } # iac/prerequisites/versions.tf aws = { source = "hashicorp/aws" version = "~> 3.66.0" required_version = "~> 1.0.11" is like package.json in node.js or go.mod in GO. Here I can specify and lock the version of provider used in the stack as well as define a version of terraform that's required to build and deploy my infrastructure. versions.tf With all that in place I could try to deploy prerequisites stack to AWS, but there's a problem. Terraform will not have access to AWS api. To solve it, I need to export secret and access keys that I got from AWS when my user with programmatic access was created: # replace ******** with appropriate values export AWS_SECRET_ACCESS_KEY =******** export AWS_ACCESS_KEY_ID =******** Time for some action: terraform init terraform plan terraform apply -auto-approve cd iac/prerequisites # initialise terraform project # plan deployment, should output: # Plan: 2 to add, 0 to change, 0 to destroy. # deploy, should output: # Apply complete! Resources: 2 added, 0 changed, 0 destroyed. One important thing to mention is that my user has full access rights to S3, DynamoDB, API Gateway and Lambda attached via AWS management console. I find it a bit easier to manage this way. If you'll run into errors, say then get back to AWS console and ensure that your user has appropriate permissions attached. not authorized to perform: dynamodb:CreateTable After running above commands, it's always a good idea to navigate back to AWS management console and ensure all resources have been created. It's also a good idea to destroy and recreate the stack to make sure everything is in right order: terraform plan # destroy the stack # should print out: Destroy complete! Resources: 2 destroyed. terraform destroy - auto -approve # plan the stack # deploy the stack terraform apply - auto -approve Sometimes terraform fails to execute. Most of the time the solution is just to rerun the last terraform command. If that does not help then your favourite search engine is always one click away and I'm pretty sure there will be plenty of answers on how to solve the issue. I hope you have learned something useful. In part 2 I'll be creating configuration for api infrastructure. This is where above prerequisites backend will be used. You can find the source code and track the progress of the project . Got inspired? Share and inspire others! here

Different

2021 - HackerNoon Contributor of the Year - GO

2022 - HackerNoon Contributor of the Year - Aws

2022 - HackerNoon Contributor of the Year - Rest Api

HELLO WORLD_

Nominated for 2022 - HackerNoon Contributor of the Year - Rest Api

Nominated for 2022 - HackerNoon Contributor of the Year - Aws

Too Long; Didn't Read

Learn how to continuously improve your codebase

Serverless API with Terraform, GO and AWS, Part 1

Too Long; Didn't Read

Company Mentioned

@danstenger

RELATED STORIES