Deploying Dockerized MLFLOW Application Server on AWS With Terraform

This project focuses on deploying containerized mlflow app on AWS provisioned through terraform. The architecture designed and executed is given below

This infra consists of a standalone networking stack including a VPC, Public and Private Subnets and other dependent resources. Furthermore, it includes an EC2 instance running the MLFlow application bootstrapped on it once instance is up and running. It also contains a custom security group attached with the EC2 Instance.

Prequisites

AWS CLI installed and configured to your AWS account
Terraform Installed

Accessing the application

You will hit the EC2 instance's public IP to view the app running on it.

<MLFlow_EC2_IP:80> - This will populate for you as an output when the apply is complete. Place into Browser.

To access the app you need to put the username and password which are given below

username: mlflow
password: mlflow

Deploying a fresh Terraform script on AWS

Follow the steps given below to deploy a fresh end to end infra provisioned by terraform

Terraform Initialization

Open the Terraform Main Folder in the VSCode and in the root directory [ which contains other directories like modules, template etc. ] ,

type the command given below for terraform initialization:

terraform init

Terraform Apply

Once terraform is initialized, run the command below to see which resources would get deployed if terraform script runs successfully:

terraform plan

The variable configuratiom files is terraform.auto.tfvars and you can change the attribute paramaters value over there of your own choice.

To deploy infrastructure, run the command below in in root folder

terraform apply

After the command starts to run, terraform will show resources being created and will ask permission from you to build up resources.

Type -- yes -- and AWS infrastructure will start to build up and get deployed.

Terraform Destroy

If you want to destroy the terraform, type the command below:

terraform destroy

On running this, terraform will show resources being destroyed and will ask permission from you to clean up resources.

Type -- yes -- and AWS infrastructure will start to clean up and resources will be destroyed.

NOTE for the sake of this PoC I have left out a gitignore to avoid any potential issues. I understand it is best practice in production