In this project, we're going to try to understand more about how computers talk to each other over a network. To do that, we're going to actually build a handful of very simple internet servers and connect to them using tools that come with our computers.

Look at the Iterations page to see how to get started.

The idea of "writing a network server in Ruby" probably seems crazy, but our first server will be about 20 lines long (excluding comments). This is a real-deal server. You could put it on an Amazon EC2 instance, run it, and any computer in the world would be able to connect to it, if that's what you wanted.

Server software is no more complicated than any other kind of software. In fact, we'll be re-designing some of our command-line software to run remotely and you'll see how little "re-design" is actually needed.

The most common type of computer networking involves one computer acting as a "client" and another acting as a "server." The server is running some software that the client wants to make use of. It connects to the server, uses the software (as the server permits), and then either ends the connection itself or waits for the server to end the connection.

One of the easiet-to-understand examples is a file server. There's a computer somewhere with a bunch of files on it. Other computers connect to it in order to access those files. Perhaps the file server allows users to copy files to their own computers but not delete them from the server itself.

Servers listen for incoming connections while clients make outbound connections.

To connect to a service running on another computer, you need two pieces of information: a host name (or IP address) and a port number.

The host name (or IP address) identifies the computer on the internet, much like a phone number does on a telephone network. The port number is an extra bit of information which tells the remote computer which service we want to use, much like a telephone extension.

We need this extra piece of information because a single computer can be running multiple pieces of server software, e.g., a web server, a file server, a mail server, and so on. Each server runs on a single port, so that the remote computer knows which piece of server software is responsible for handling the client.

Host names look like "cnn.com", "facebook.com", and so on. IP addresses look like "74.125.239.38", "173.252.110.27", and so on. Port numbers are just plain numbers. If you've ever seen a URL that looks like

www.google.com:1234

That 1234 is a port number.

Software engineers use the word "server" to mean two different things:

1. The computer on which server software is running
2. The server software itself

To a computer (the client) trying to connect to a remote service, the distinction doesn't much matter, which is why engineers use the word to mean both things in casual conversation.

Likewise, when we say "client" we sometimes mean the software making the connection (e.g., your web browser) or the computer on which that software is running.

Since we'll be running our server software on our own computers, we'll almost always mean "server software" when we say "server." We'll let you know when the distinction is important.