Why I Don’t Teach The OSI Model

2022 Update: If you want a clear technical explanation of the OSI model, click on the cover of my book CCNP Enterprise Certification Study Guide: Implementing and Operating Cisco Enterprise Network Core Technologies for a free preview of Chapter 1 which explains each layer of the OSI model in intricate detail.

I recently got an email from a viewer of my Practical Networking course who asked how the TCP/IP networking terms I used mapped to the Open Systems Interconnect (OSI) model.

First, a bit of background. The OSI model is a generic networking model that is supposed to describe conceptually how networks carry data. Within the last four decades or so, 99.9% of all computer networking curricula for beginners has started by rehashing the OSI model.

When I first started out learning networking, I paid my dues by memorizing the 7 layers of the OSI model: application, presentation, session, transport, network, data link, and physical. But I found it almost useless in understanding how modern TCP/IP networks actually work.

When I began teaching networking, I found that it was clearer to simply explain things without ever explaining the OSI model. It’s an approach that’s worked well, as evidenced by the many compliments I’ve gotten on my networking courses and books.

The sad fact is that you don’t need to know the OSI model. All you need to know is how people use the terms. Here you go:

Layer 1 – Physical

The electrical signaling, physical connections, the bits. “We have a layer 1 problem” sometimes means “a rat chewed through the cable” or “it’s raining and the humidity is attenuating the signal.”

Layer 2 – Data link

Ethernet technologies, including MAC addresses, Ethernet frames, VLANs and VLAN tags; serial encapsulation such as the point-to-point protocol (PPP). Much of the time “the problem is at layer 2” means “it’s in the wrong VLAN”.

Layer 3 – Network

IP addressing, IP routing, and address resolution protocol (ARP); IPv6, neighbor discovery (ND), and the like. “We have a layer 3 problem” can mean “we have a routing problem” or “someone put in the wrong IP address.”

Layer 4 – Transport

Transmission control protocol (TCP) and User datagram protocol (UDP); This includes TCP and UDP port numbers. Incidentally, few people use this in conversation. Instead, they say “layer 7” when they mean layer 4, which brings us to…

Layer 7 – Application

Technically, this is just the data payload that the network carries. Strangely, in troubleshooting conversations, “a layer 7 problem” often means “a firewall is blocking that port”, referring to a TCP or UDP port number, distinctly a layer 4 problem. The confusion arises from the fact that most standard applications have a registered port number they use. For example, TCP port 80 is for the HTTP application, so people use the two interchangeably.

What about the other layers?

Nobody uses them. Seriously. In TCP/IP networks, session and presentation are rolled up into the application layer, which is itself just the data that you’re sending across the network. In fact, when you think about it, it makes perfect sense. What’s the point of a network? To transport data. What’s the highest layer that actually is part of the network infrastructure? That’s right, the transport layer.