The TCP/IP Model and Encapsulation

The OSI model is the classroom map; the TCP/IP model is the one the internet was actually built on. Kite Studio's router manual, Linux networking commands and most vendor docs speak in TCP/IP terms, so you need to translate between the two fluently — and to see how every layer wraps the one above it.

The rule

Four layers, each wrapping the one above: headers on the way down, off on the way up.

Some books split Network Access into separate Data Link and Physical layers and call it a five-layer model; the idea is the same.

Encapsulation is how the layers cooperate. On the sending side, each layer takes what the layer above handed it, treats it as an opaque payload, and adds its own header in front:

1. Application: the HTTP request itself — the data.
2. Transport: adds a TCP header with source and destination ports → a segment.
3. Internet: adds an IP header with source and destination IP → a packet.
4. Network Access: adds an Ethernet header with MACs and a trailer with a checksum → a frame, sent as bits.

The receiver does the reverse — decapsulation — each layer reading and removing only its own header, then passing the rest up. No layer needs to understand the others' headers.

The shortcut

Picture nested envelopes, and remember the order from the inside out: Data → Segment → Packet → Frame. For any header field, ask "which envelope is it written on?": ports on the segment, IPs on the packet, MACs on the frame.

The trap is thinking a router opens every envelope. A router works at the Internet layer: it swaps the frame and reads the IP header, but leaves the segment and the data inside untouched (address translation, coming in the IP Addressing section, is the one exception).

Drills

Answer fast — trust the rule.

Remember

• Four layers — Application, Transport, Internet, Network Access — each wrapping the one above.
• Data → Segment → Packet → Frame: ports, then IPs, then MACs.
• Each device opens only as many envelopes as its job needs; a router stops at the IP header.