Skip to content

The Truth about the OSI Model

    The Truth about the OSI Model

    In a recent video discussing the “forgotten layers” of the OSI model (layers 5, 6, and 7), and my previous two videos on the OSI model, I referred to various “responsibilities” associated with each layer of the OSI model:

    • Layer 1 – Physical – Transporting Bits
    • Layer 2 – Data Link – Hop to Hop
    • Layer 3 – Network – End to End
    • Layer 4 – Transport – Service to Service
    • Layer 5 – Session – Identifying User sessions
    • Layer 6 – Presentation – Interpreting 1s and 0s
    • Layer 7 – Application – Application Commands

    The truth is, all of that was a bit of a simplification. I intentionally used this simplification to teach how Networking works and how data moves through a network.

    In reality, what I described above about the OSI model is not entirely accurate. The OSI model has never been widely used in mainstream networking — not even back in the 70s when the OSI model was invented.

    The real story of the OSI Model

    The OSI model was initially meant to be the definitive model for how Networking works. But the development of this model was hindered by excessive input and guidance by committee… which led to a “too many cooks in the kitchen” situation and a finalized model that never came to fruition.

    As the debates and discussions around what should be included in the OSI model were underway, a much simpler “straight to production” model was created — the TCP/IP model. As time went on, with the industry awaiting the finalized OSI model, more vendors started building devices and software to the specifications of the TCP/IP model. Eventually the industry lost faith that a functional OSI model would ever come to fruition and the TCP/IP model became the de facto networking stack.

    This is why when you configure a Windows machine with an IP address, you are modifying “TCP/IP” settings.

    The complete story of the OSI model (and its demise) is rather fascinating, and you can read about it here:

    https://spectrum.ieee.org/osi-the-internet-that-wasnt

    Learning Networking: The Role of the OSI Model in 2024

    But where does that leave you? The diligent learner trying to synthesize the knowledge of how Networking works.

    Well, for better or for worse, the OSI model in education appears to be here to stay. And as an educational tool, if taught correctly, I think it can fare well enough (for proof, look no further than the abundance of comments on my videos attesting to how my approach made networking “click”).

    The key is to understand that not everything fits neatly into the layered approach. And that each responsibility that I went through in my training videos, is not necessarily confined to the layer, or the sequence, in which they were described.

    For example, in my video on Layers 5, 6, and 7, I described the Session Layer (L5) using HTTP cookies as an example of “identifying a user session“. But you can’t read a cookie to identify a user session until the application has already “interpreted the 1s and 0s” — which I described as a Presentation Layer (L6) responsibility. So clearly, in the case of HTTP, the responsibility of “interpreting the 1s and 0s” has to occur before the responsibility of “Session to Session delivery”.

    So you see, the responsibilities themselves matter, but their strict association to a particular layer is less important.

    Instead, I would encourage you to learn the OSI model from the perspective of “layers of abstraction. Something has to handle the end to end delivery of packets and IP addresses and routing. Something has to handle disambiguating incoming network streams. And if you want network protocol mobility, something has to exist to identify your session independent of hardware or location specific addresses (i.e., MAC addresses and IP addresses).

    Transitioning from the OSI model to the TCP/IP model

    The strategy of teaching “responsibilities” within each “layer of abstraction” could absolutely be taught with equal effectiveness using the more modern 5-layer TCP/IP model. I am a proponent of this approach, as it’s a simpler, streamlined model for understanding computer networking, without the historical baggage of the OSI model.

    But alas, much of the Computer Networking educational content and certification requirements still use the OSI model as a foundation. So, as mentioned earlier, for better or worse, the OSI model appears to be here to stay.

    I am encouraged to see some of the recent Cisco CCNA curriculum make more references to the 5-layer TCP/IP stack. Yet, the inclusion feels tentative; leading to, in my opinion, confusion between the two models. Perhaps a more decisive shift towards the TCP/IP model is a more effective approach.

    But of course, for the sake of creating well rounded candidates, every Networking certification has to include the basics of what every other curriculum teaches, so it’s possible we are in a stalemate — entrenched with the OSI model forever.

    The day all of Computer Networking education (and troubleshooting, and Network Engineering day-to-day lingo, and so on) transition fully to using the TCP/IP model, I’ll happily recreate my videos with a focus on the 5-layer TCP/IP model.

    Until then, I am still a proponent of teaching the OSI model, so long as it is taught the right way: Focusing on the practical aspects necessary for understanding how data moves through a network.

    There is no shortage of terrible educational content on the OSI model — full of rote memorization and incoherent definitions and half-truths. To see examples of these, look no further than Robert Graham’s Twitter thread, or his 200-page exposition unpacking the extent to which the OSI model is a lie.

    5 11 votes
    Article Rating
    Subscribe
    Notify of

    3 Comments
    Oldest
    Newest Most Voted
    Inline Feedbacks
    View all comments

    well written article thank you

    thanks bro share us

    Very nice articles. Thank you very much for sharing!