The OSI Model as a Map, and One Device or All?
Where you are: Module 9 — Network Troubleshooting · Lesson 9.1 · Based on Chapter 9 (introduction, Figure 9.1 and §9.1) of Dr. Tahseen Al Doori's Troubleshooting Essentials.
What you'll be able to do
- Name the seven layers of the OSI model and give one troubleshooting check for each.
- Explain why a fault at a lower layer produces symptoms at every layer above it.
- Use the question "one device or all?" to decide whether a fault lies with a single device or with shared equipment.
- Choose a sensible starting point (bottom-up, top-down or the middle of the stack) for a given network symptom.
Watch: The OSI Model as a Map, and One Device or All?
The video follows a fictional support ticket, "the internet is down", from a finance office. It climbs the OSI model layer by layer, shows how a single unplugged cable produces a browser error, and then proves the cause with three quick tests. As you watch, ask yourself:
- Where does the user see the problem, and where does the fault actually live?
- Which single question removes half of the possible causes before any test is run?
- Why is a ping to the default gateway such a useful first test?
Why networks need a map
Network faults are among the most frequent and most disruptive problems a support technician meets. They are also confusing, because the symptom rarely names the cause. "The internet is down", "Teams won't connect" and "the website is broken" can all come from the same loose cable.
The book's answer is to use the OSI model as a map. OSI stands for Open Systems Interconnection: a reference model that describes network communication in seven layers, from the physical cable at the bottom to the application the user sees at the top. You do not need to memorise every protocol to use it. For troubleshooting, one idea matters most:
Each layer depends on the layers beneath it, so a fault at a lower layer causes symptoms at every layer above.
That single sentence gives you a natural order for testing. If the bottom of the stack is broken, nothing above it can work, however healthy it is.
The seven layers as a checklist
The table follows Figure 9.1 in the book, which pairs each layer with the things you would actually check. Terms are defined in plain language in the last column.
| Layer | What you check | In plain terms |
|---|---|---|
| 7 Application | Application errors, the browser, the email client, whether the DNS name was typed correctly | The program the user is working in |
| 6 Presentation | Encoding, encryption, certificate and TLS errors | How data is formatted and protected; TLS is the encryption behind secure websites |
| 5 Session | Logins, sessions timing out, authentication | Keeping a conversation between two systems open |
| 4 Transport | Are ports open? Is a firewall blocking TCP or UDP? (netstat) | Delivering data to the right program on the right machine |
| 3 Network | IP address, subnet mask, default gateway, routing (ipconfig, ping, tracert) | Addresses and routes between networks |
| 2 Data Link | Switch port, VLAN, Wi-Fi association, MAC address | The link to the next device; a MAC address is the network card's hardware address, a VLAN is a virtual network on a switch |
| 1 Physical | Power, cables, link lights, Wi-Fi signal, hardware | Things you can see, touch or measure |
Notice where the familiar tools sit. ping, ipconfig and tracert all work at layer 3, which is why they appear so often in this module. DNS, the service that turns names into addresses, is an application-layer protocol; the book's symptoms table therefore places name-resolution faults at the Application layer.
Symptoms appear high, faults often live low
Imagine a network cable has been pulled out. That is a layer 1 fault. The user, however, does not experience "layer 1"; they experience a browser that says the site cannot be reached, an email client that will not sync and a chat application that keeps reconnecting. Every one of those complaints is a layer 7 symptom of a layer 1 cause.
This is why the experienced technician does not start by reinstalling the browser. The symptom tells you that something below the application has failed. The layers tell you where to look.
Three places to start
The book's figure suggests three starting points. They are the same ideas you met in Module 6 (Isolation Techniques), applied to a network.
Bottom-up from layer 1
Start with power, cables and link lights. Best when hardware or connectivity is suspected, or when there is no obvious starting point.
Top-down from layer 7
Start with the application. Best when only one application misbehaves and everything else works.
Split at layer 3
Start in the middle with a ping. If it succeeds, layers 1 to 3 are working and the fault is higher; if it fails, look downwards.
The third option is the half-split method from Module 6 applied to the stack. One test removes roughly half of the possibilities.
The first question: one device or all?
Before climbing any layers, the book asks a single scoping question.
| Who is affected? | Where the fault lies | Check first |
|---|---|---|
| One device | That device, its cable or its configuration | The device's cable, Wi-Fi and IP settings |
| Every device | Shared equipment (switch, router, modem) or the internet service provider | The shared equipment and the provider's status |
This question halves the problem immediately. If twenty computers lost the internet at the same moment, it is almost impossible that twenty cables failed together; something they share has failed. If one computer is offline while its neighbours work, the shared equipment is cleared and you can concentrate on that one machine.
Answering the question usually takes a minute: ask the user whether colleagues are affected, look at a second device, or check whether the problem has been reported elsewhere. It is the network version of the "Who" question in the 5W1H set from Module 4.
Worked example
(Fictional scenario.) At 09:10 on a Tuesday, a member of the finance team reports that "the internet is down" on their wired desktop PC. Their desk was moved the previous afternoon.
- Identify. The technician asks the scoping question. Colleagues on the same floor are working normally, so this is one device, and the switch, router and provider are cleared.
- Theory. Because the desk was moved, the most likely cause is physical: a cable or connector disturbed during the move. Bottom-up is the obvious approach.
- Test. The link light on the PC's network port is off.
ipconfigreports the Ethernet adapter as Media disconnected, which means there is no physical link. A ping to the gateway cannot even be sent. Every result points to the first link, between the PC and the wall socket or switch. - Plan and implement. The cable has been pulled half out of the wall socket. It is reseated firmly; nothing else is changed.
- Verify.
ipconfignow shows a valid address, subnet mask and default gateway. A ping to the gateway (192.168.1.1 in this example) gets steady replies, and the user confirms that their usual sites and email work. - Document. The ticket records the cause (cable dislodged during a desk move) and the fix. A note is added to the office-move checklist: test every connection after furniture is moved.
Try it
Do this on your own computer or a lab machine. It is observation only; you will not change any settings.
- Write the seven OSI layers on a sheet of paper, layer 1 at the bottom.
- For your own connection, find one piece of evidence for layers 1, 2 and 3: for example the Wi-Fi signal bars or a link light (layer 1), the name of the network you have joined (layer 2), and your IP address and default gateway from
ipconfigon Windows,ifconfigon macOS orip addron Linux (layer 3). - Ping your default gateway and note whether replies come back.
- Ask one other person on the same network whether their connection is working right now, and record the answer as "one device" or "all devices" practice.
You have finished when your sheet shows all seven layers, at least three pieces of evidence placed on the correct layers, and the result of your gateway ping.
Knowledge check
Why does the OSI model give a natural order for network testing?
A user reports that a website will not load. The cause turns out to be an unplugged cable. At which layer did the fault live?
All twenty computers in an office lose internet access at the same moment. Where should you start?
Only one application misbehaves; email, browsing and file sharing all work. Which starting point fits best?
Answers explained
- a — lower layers support higher ones, so testing from a sensible starting point and following the dependencies narrows the fault; manufacturers, likelihood rules and "one layer only" are not why the model works.
- c — the cable is layer 1; the user only saw the symptom at the Application layer, and Transport and Session were working perfectly.
- b — when every device fails together, the cause is usually something they share; twenty separate cable or browser faults at once is very unlikely, and a keyboard layout has nothing to do with it.
- d — the book recommends top-down when only one application is affected; the network is plainly working for everything else, so hardware and provider checks come later, if at all.
Key takeaways
- The OSI model is a troubleshooting map: each layer depends on the layers beneath it.
- Symptoms usually appear at the application layer, even when the fault is physical.
- Ask "one device or all?" first; it halves the problem before any test.
- Start bottom-up for hardware faults, top-down for a single application, or at layer 3 with a ping to split the stack in half.
Sources
- Dr. Tahseen Al Doori — Troubleshooting Essentials, University Edition (2026), Chapter 9 introduction, Figure 9.1 and §9.1; §6.1 and §6.4 (isolation techniques)
