The Network Stack & Protocols

The internet is not one program — it is a stack of layers, each doing one job and trusting the layer below to do its own. This separation is the reason the internet works at all: Wi-Fi can change to fibre, or HTTP to a video protocol, without anything else needing to know. Each layer talks only to the layers directly above and below it.

The engineering idea is encapsulation: as your data travels down the stack to be sent, each layer wraps it in its own header (like putting a letter in an envelope, then a mailbag, then a truck); on the way up at the other end, each layer unwraps its own part.

Follow a message down the sending stack and back up the receiving one. Watch each layer add its header on the way down and strip it on the way up — that is encapsulation.

Two models are taught. The TCP/IP model (4 layers) is what the internet actually runs; the OSI model (7 layers) is the fuller teaching reference. Mapped to what each does:

The transport layer offers two very different services, and choosing between them is a classic question. TCP is reliable: it sets up a connection, numbers every segment, resends anything lost, and delivers everything in order — at the cost of speed and overhead. UDP is a fire-and-forget datagram: no connection, no resending, no ordering — but fast and lightweight.

So downloads, web pages and email use TCP (you need every byte, correct and in order); live video, voice calls and online games use UDP (a lost frame is better dropped than delivered late).

One computer runs many network programs at once, so the transport layer uses port numbers to say which program a message is for — like flat numbers within a building whose street address is the IP. Web servers listen on port 80 (HTTP) and 443 (HTTPS); DNS on 53.

HTTP is the application protocol of the web — a request for a page, a response with it. HTTPS is the same wrapped in TLS encryption, so nobody between you and the site can read or tamper with the traffic. It is why the padlock matters.