Don’t Panic: How the Internet Works

The Internet is the arrangement that lets independently run networks pass packets to one another. It is the plumbing beneath the Web, email, calls, and APIs. Calling it the Web is like calling a railway system a timetable. The timetable is useful, but it is not what moves the trains.

A request begins with a domain name, such as example.com. DNS turns that human-friendly label into records that can guide a browser toward an address. The browser then sends IP packets through routers. Each router makes a next-hop decision, which is pleasingly mundane for something that appears to involve a globe and several satellites. The route may cross many networks, and no one operator owns all of them.

The packet layer is deliberately modest. IP tries to move packets toward an address. It does not promise that they arrive, arrive once, arrive in order, or arrive wearing a tiny security hat. TCP adds ordered and reliable byte delivery. QUIC provides reliable streams over UDP. TLS protects a connection to the endpoint it authenticates. HTTP gives the exchange its web meaning: a request, then a response.

This separation is the useful bit. A domain name problem is not a routing problem. A routing problem is not a TLS certificate problem. An HTTP error means the request got surprisingly far, but it does not prove the application did the right thing. When a page feels slow, split the timing into DNS, connection, TLS, first byte, and transfer. The total is a mystery novel; the stages are the clues.

Start with the slides for the map of layers and boundaries. Use the cheatsheet when a name, address, port, packet, or route begins impersonating another one. The practice reference gives you a safe way to observe a public request, and the quiz checks whether the layers have stopped swapping hats.