This is an example of a fully scaleble api deployment on ECS Fargate.

Stack used

AWS:

• ECS - Fargate
• ALB
• And all supportive service (IAM, Security Groups, EC2, etc.)

Shell Script

Python

Docker

Terraform

Some ideas that could be improved over time

• Nginx as a reverse proxy inside the container;
• Terraform modules instead of a single project resources;
• Terraform state would be remote;
• Substitute build.sh script for some configuration tool like Ansible that would be easier to read and maintain the code;
• Substitute cloudwatch standard monitring with different monitoring stack;
• Use actions for deployment instead of running shell commands;

Deploy Instructions:

DO NOT FORGET TO CONFIGURE YOUR AWS CREDENTIALS AS YOU WISH. I ADVISE TO USE:

aws configure

Add to your Github global variables your AWS credentials: aws_access_key_id and aws_secret_access_key.

Change aws_acc_id and AWS_ACC_ID values on main.tf and build.sh respectively, to your AWS's account id.

Manually create a ECR repository named service on the AWS's region you've chosen to deploy you application. The region value must be specified on main.tf and build.sh.

Container image should be built and pushed to ECR as soon as a commit goes to master and everything above is set up. In case it doesn't work, just run the script deploy.sh with pipeline option first, it will build and push the container image to ECR.

./build.sh pipeline

Run the script deploy.sh, it will do the whole job, including the installation of the tools needed.

./build.sh deploy

Destroy Instructions:

Run the script destroy.sh, it will do the whole job.

./build.sh destroy