Host a Static Website with Nginx

This is the payoff page for everything in the EC2 section. You'll take a blank instance and turn it into a real, public web server - using the security groups, IPs, and Nginx pieces you've met along the way. By the end, you type your instance's address into a browser and your own page loads.

What we're building

   Browser  ──HTTP (port 80)──►  Security Group (allows 80)
                                       │
                                       ▼
                                 EC2 instance
                                 └── Nginx serving /usr/share/nginx/html

Nginx is a fast, lightweight web server. Its job here is simple: listen on port 80 and hand back HTML files from a folder on disk.

The prerequisites checklist

Before the website can work, three things must be true - and each maps to a page you've already read:

If port 80 isn't open, the site is invisible no matter how perfectly Nginx is configured. This is the #1 reason "it works locally on the server with curl but not in my browser."

Step by step

ssh -i my-key.pem ec2-user@<public-ip>
sudo yum update -y
sudo yum install -y nginx        # AL2023: sudo dnf install -y nginx

sudo systemctl start nginx
sudo systemctl enable nginx      # so it survives reboots
sudo systemctl status nginx      # should say "active (running)"

echo "<h1>My first AWS-hosted site</h1>" | sudo tee /usr/share/nginx/html/index.html

Do it all automatically with user data

Everything above can be the instance's user-data script, so a fresh instance comes up already serving your site - no SSH at all:

#!/bin/bash
yum update -y
yum install -y nginx
systemctl enable nginx
systemctl start nginx
echo "<h1>Served automatically on first boot</h1>" > /usr/share/nginx/html/index.html

Paste that into Advanced details → User data at launch, open port 80, and the site is live the moment the instance is ready.

When it doesn't work - the debugging ladder

Connection problems almost always fall into one of these, in order of likelihood:

That curl localhost test is the key diagnostic: it splits the problem cleanly into "is Nginx serving?" (server side) versus "can the world reach it?" (network side).

Where to go from here

A single instance serving HTML is a real website, but it has real limits - and each limit points to a later topic:

That wraps up compute. You can launch a server, secure it, give it storage, back it up, bootstrap it, and serve a site from it. Next we stop babying a single instance and make a fleet resilient - with load balancing and auto scaling.