AWS Infrastructure Project

Overview

This project builds the infrastructure for a media-heavy blog platform — one where users upload images and video, and traffic can spike unpredictably. Rather than a single EC2 instance, the goal was a setup that scales horizontally and stays available under load.

Infrastructure layout

Everything lives inside a dedicated VPC:

• Application Load Balancer (ALB) — distributes incoming traffic across instances and terminates TLS
• Auto Scaling Group of EC2 instances — scales out on CPU/traffic thresholds, scales back in when load drops
• RDS — a managed relational database, isolated in private subnets, not reachable from the public internet
• S3 — stores uploaded images and videos, decoupling media storage from the application servers entirely
• CloudFront — CDN in front of S3 and the ALB, caching static assets close to users
• Route 53 — DNS and health-check based routing in front of CloudFront

Why offload media to S3

Storing uploads on the EC2 instances themselves would tie media to a single machine — a scaling nightmare and a single point of failure. Writing uploads directly to S3 (via presigned URLs from the backend) means:

• Any instance in the Auto Scaling Group can serve any request, statelessly
• Storage scales independently of compute
• CloudFront can cache media at edge locations without touching the origin servers

What I'd improve next

• Move RDS to a Multi-AZ deployment for automatic failover
• Add a WAF in front of CloudFront for basic request filtering
• Script the whole stack with Terraform instead of manual console setup, for repeatability