A Very Brief History

In the early days of the world wide web — before IPv4 address exhaustion was a concern — a web site would be running on a single IP address. So, one IP address per website. Not only is this a concern for IPv4 exhaustion, but it's also not a very efficient use of server infrastructure and resources.

As the Internet — and the web by extension — became more commoditized, more individuals were setting up websites to share information with the world. Also among the mix, were hosting providers giving space for people to create a website without the need for a personal server of their own.

The idea now is that you'd have many websites on a single IP address. But, this begs the question, "How is this possible?". The answer is virtual hosting and it's not as complicated as it sounds.

Many Websites, One IP Address

The answer lies in the fundamentals of TCP/IP networking and where things happen at different layers.

• IP-based virtual hosting — One host has many IP addresses
• Port-based virtual hosting — One TCP port maps to one website
• Name-based virtual hosting — Relies on application layer protocols
  • DNS
  • HTTP

In this post, we are going to be focusing name-based virtual hosting, as this is the version most in-use today.

Virtual Hosting Diagram

Making Sense of the Diagram

1) SSH into Web Server and Create the Virtual Hosts

The IT administrator for Example Corp. logs into the web server and modifies the configuration for their web server. For this example, we'll say they're running Nginx.

example.com and www.example.com

sudo nano /etc/nginx/sites-available/example.com.conf

server {
   listen 80;
   server_name example.com www.example.com;
   root /var/www/website;
   index index.html;

   location / {
       try_files $uri $uri/ =404;
   }
}

sudo ln -s /etc/nginx/sites-available/example.com.conf /etc/nginx/sites-enabled/example.com.conf

blog.example.com

sudo nano /etc/nginx/sites-available/blog.example.com.conf

server {
   listen 80;
   server_name blog.example.com;
   root /var/www/blog;
   index index.html;

   location / {
       try_files $uri $uri/ =404;
   }
}

sudo ln -s /etc/nginx/sites-available/blog.example.com.conf /etc/nginx/sites-enabled/blog.example.com.conf

Restart Nginx

nginx -t && sudo systemctl restart nginx

2) Add the DNS Records

The IT administrator logs into the DNS provider for Example Corp. and modifies the example.com DNS zone.

1. @ — zone root, points at the IP address of their web server
2. www.example.com — points at the IP address of their web server
3. blog.example.com — points at the IP address of their web server

The DNS records are published and should resolve within a matter of seconds to minutes.

3) Client Navigates to Site

example.com or www.example.com

1. User opens their web browser and navigates to example.com
2. Computer makes a DNS query for example.com and receives the IP from the name server
3. Format the HTTP Request

GET / HTTP/1.1
Host: example.com
Header: value
Header: value

4. TCP/IP Networking

• The HTTP application data is encapsulated down into the transport layer
• The transport layer is encapsualted down into the network layer
• The network layer is transported down into the Ethernet frame
• The packet is put on the wire and transported over many intermediate routers
• The packet reaches the web server for example.com

5. Web Server processes request

• The web server's network interface decapsulates the packet
• The destination port is tcp/80 and is passed on to Nginx
• Nginx parses the HTTP data
• The HTTP data contains Host: example.com
• A valid configuration exists for this hostname at /etc/nginx/sites-enabled/example.com.conf
• The server loads /var/www/website/index.html as the site root is defined as /var/www/website

6. TCP/IP Networking

• The web server follows the same steps as mentioned above
• Returning the index.html data back to the client

blog.example.com

The process is exactly the same as above except:

1. The user opens their web browser and navigates to blog.example.com
2. The computer makes a DNS query for blog.example.com and gets the IP address from the name server
3. The HTTP request now contains the Host: blog.example.com header

GET / HTTP/1.1
Host: blog.example.com
Header: value
Header: value

4. When the Nginx server receives the request, the Host: blog.example.com header matches the configuration in /etc/nginx/sites-enabled/blog.example.com.conf with server_name blog.example.com; directive
5. The Nginx server returns /var/www/blog/index.html to the user

What About the Stage Hostname?

You may have noticed in the web server, there is another virtual host for stage.example.com. But, this virtual host does not have a DNS record published in the example.com domain.

That begs the question, "If the virtual host exists, but the DNS record does not, how can people navigate to it in their web browsers?"

There are two possible solutions to this:

✅ Example Corp. has an internal DNS server which resolves the hostname for people working in the office or connected to the VPN.

✅ A user could modify their hosts file to create a local DNS name if there are no DNS servers able to resolve the DNS query.

• On Linux the hosts file is found at /etc/hosts
• On Windows, it's found at C:\Windows\System32\etc\drivers\hosts

sudo nano /etc/hosts

88.131.47.75        stage.example.com

curl -k https://stage.example.com