Day 15: OSI as a mental map
The spine of this phase
This phase's capstone (Day 23) is narrating everything that happens when you type google.com and hit enter, packet-by-packet. Everything from here to there builds toward being able to answer that in real interview depth. The OSI model is the map we'll use to organize it — not because real systems obey it perfectly, but because it gives every concept a slot.
- Physical — actual bits on a wire or radio signal
- Data Link — frames between devices on the same local network (Ethernet, MAC addresses)
- Network — routing packets between networks (IP)
- Transport — reliable or unreliable delivery between processes (TCP, UDP)
- Session — managing a conversation over time
- Presentation — data format/encoding (TLS often gets loosely placed here)
- Application — what your program actually talks (HTTP, DNS)
Nested envelopes
Your HTTP request is an envelope. TCP wraps it in another envelope (adding sequencing info). IP wraps that in another (adding source/destination addresses). Ethernet wraps that in one more (addressing the next physical hop). Each layer only reads its own envelope — the layer below doesn't care what's inside.
Why bother with a 7-layer model most systems don't cleanly follow
Interviewers use "what layer does X operate at?" as a fast way to check whether you actually understand a technology's job, or just its name. A load balancer working at "L4" vs "L7" (Day 23) is a real, consequential distinction, not trivia.
Key terms
- OSI model
- A 7-layer conceptual map of networking responsibilities, from physical bits to application data.
- Encapsulation
- Each layer wrapping the layer above's data in its own header/envelope before passing it down.
When people say a load balancer operates "at L4" vs "at L7", what are they distinguishing?
