I installed cdk and ran cdk init selecting python as my chosen language.
In order to get stated, I consulted this repo:
https://github.com/aws-samples/aws-cdk-examples
for inspiration.
In order to use 3 AZs, a specific and supported region is required to be specifed in app.py
The Vpc construct includes the creation of subnets, routes, IGW and NATs.
We configured three different subnets (per AZ):
- Public
- Private
- DB
I created one Application Load Balancer.
I created an Amazon Aurora PostGres RDS database ensuring that it used the DB Subnets. The cluster creates a DB instance per AZ.
I created 6 simple EC2 instances using the default Amazon Linux 2 AMI. I had Apache httpd installed on initial startup using user-data, and a simple Hello World webpage is installed. This is not available publically.
I created an ECS Cluster and an auto-scaling group and then configured the ASG to be the ECS Cluster's Capacity provider. If further services are added then the ASG would provision more EC2 instances.
I chose to use the sonarqube:8.9.8-community image in an EC2 service in this cluster, and configured it to use the PostgresSQL RDS cluster as a DB backend that I configured earlier.
After initial deployment, the default username and password of admin and admin are used to login to the web application.
This can be accessed by surfing to the LB's public DNS name under port 8080 (as this is where the respective listener is configured)
I created a simple ECS Cluster with no such capacity provider.
For this example I used a sample amazon/amazon-ecs-sample image and ran it with Fargate (instead of EC2).
This can be accessed by surfing to the LB's public DNS name under port 8081 (as this is where the respective listener is configured)
I created a simple empty S3 bucket.
I was able to deploy the stack twice by instantiating another MyProjectStack in app.py
It is worth noting that, by default, an AWS account is limited to just 5 Elastic IPs. This can be requesting it of AWS.
Cross-region and regular backups could be taken of the RDS Database. In the event of disaster. The deployment could be moved to a different region.