This project was submitted as a final project for the Infrastructure As Code (IAC) topic of Udacity's Cloud Devops Engineer Nanodegree.
A realistic scenario requiring one to deploy a dummy application (a sample JavaScript or HTML file) to the Apache Web Server running on an EC2 instance was chosen for this project.
- AWS account (& AWS CLI)
- A code editor, such as VS Code
- Lucidchart account
Your company is creating an Instagram clone called Udagram.
Developers want to deploy a new application to the AWS infrastructure.
You have been tasked with provisioning the required infrastructure and deploying a dummy application, along with the necessary supporting software.
-
You'll need to create a Launch Configuration for your application servers in order to deploy four servers, two located in each of your private subnets. The launch configuration will be used by an auto-scaling group.
-
You'll need two vCPUs and at least 4GB of RAM. The Operating System to be used is Ubuntu 18. So, choose an Instance size and Machine Image (AMI) that best fits this spec.
-
Be sure to allocate at least 10GB of disk space so that you don't run into issues.
-
Since you will be downloading the application archive from an S3 Bucket, you'll need to create an IAM Role that allows your instances to use the S3 Service.
-
Udagram communicates on the default
HTTP Port: 80
, so your servers will need this inbound port open since you will use it with the Load Balancer and the Load Balancer Health Check. As for outbound, the servers will need unrestricted internet access to be able to download and update their software. -
The load balancer should allow all public traffic
(0.0.0.0/0)
onport 80
inbound, which is the default HTTP port. Outbound, it will only be usingport 80
to reach the internal servers. -
The application needs to be deployed into private subnets with a Load Balancer located in a public subnet.
-
One of the output exports of the CloudFormation script should be the public URL of the LoadBalancer. Bonus points if you add
http://
in front of the load balancer DNS Name in the output, for convenience.
Taking the project guidelines and the project rubric into consideration, I came up with the following infrastructure diagram for the deployment of Udagram:
This image was interpreted into a CloudFormation template script and Parameters script.
The create.sh script was used to create stacks. Usage:
~$ ./create.sh myStack myStackTemplate.yml myStack-params.json
Where:
myStack
is the name of the stack to be createdmyStackTemplate.yml
is the template script for the stackmyStack-params.json
is the parameters file
Similarly, the update.sh script can be used to update an existing stack:
~$ ./update.sh myStack myStackTemplate.yml myStack-params.json
The delete.sh script deletes the stack. The name of the stack is passed to it:
~$ ./delete.sh myStack
The screenshot below showcases successful deployment of a dummy website by the scripts.
The resources were then successfully deleted after the success of the exercise. (Update: The above URL is invalid now).