How to troubleshoot a network (OSI Layers 1–7)
The fast answer: start at the bottom and climb
The reliable way to troubleshoot any network problem is to work through the OSI model from the bottom up — start at Layer 1 (the physical cable and link), confirm it's solid, then climb one layer at a time until you find the layer where things break. This stops you from rebooting routers and reinstalling drivers when the real problem was a loose cable. The OSI model is just a seven-rung ladder that describes everything a network does, from electrical signals on a wire up to the app on your screen. Troubleshoot it like a ladder: check each rung in order, and the first broken rung is your fault.
📶 Network troubleshooting series — OSI Layers 1–7
What the OSI model actually is
OSI stands for Open Systems Interconnection. It's a conceptual model that breaks networking into seven layers, each handling one job and handing off to the next. You don't need to memorize the theory — you just need to know that a fault lives on one specific layer, and naming that layer tells you exactly what to fix. Here are the seven, bottom to top:
- Layer 1 — Physical. The cable, port, NIC, radio signal — the actual electrical or wireless link.
- Layer 2 — Data Link. MAC addresses, Ethernet frames, switches and VLANs — how devices talk on the local segment.
- Layer 3 — Network. IP addresses, routing, gateways — how packets find their way across networks.
- Layer 4 — Transport. TCP and UDP, ports, firewalls — how a reliable conversation is set up and kept ordered.
- Layer 5 — Session. Opening, maintaining and tearing down a connection between two applications.
- Layer 6 — Presentation. Encryption (TLS), compression and data formatting — the certificate handshake lives here.
- Layer 7 — Application. The thing you actually use: the browser, DNS, the web server, the app's own logic.
Why you troubleshoot bottom-up
Every layer depends on the one below it. Layer 3 routing can't work if Layer 2 can't pass a frame, and Layer 2 can't pass a frame if Layer 1 has no link light. So if you start at the top — say, blaming the website or the app — you can waste an hour on a problem that a $5 cable swap would have fixed. Starting at the bottom means the first thing you confirm is the cheapest and most common cause, and you never debug a higher layer while a lower one is silently broken underneath it. Cables and connectors are, by a wide margin, the most frequent culprit, which is exactly why Layer 1 goes first.
The divide-and-conquer method
You don't always have to climb one rung at a time. A faster variation is to test the middle and split the problem in half. A classic split: can you ping your own router or gateway? If yes, Layers 1 through 3 inside your home are healthy and the fault is higher up or further out — skip straight to DNS, the app, or the internet path. If no, the fault is below Layer 3 and local — check the cable, the link light, your IP address and the switch. Each test you run should cut the list of possible layers roughly in half until one layer is left standing.
Symptom → likely layer
Use this table to jump to the right starting layer. Match the symptom, then open that part's guide.
| Symptom | Likely layer |
|---|---|
| No link light, cable feels loose, port dead | Layer 1 — Physical |
| Wi-Fi very weak or drops in and out | Layer 1 — Physical |
| Reach some local devices but not others; VLAN isolation | Layer 2 — Data Link |
| No IP address, wrong subnet, can't reach the gateway | Layer 3 — Network |
| Connection refused / blocked port / firewall | Layer 4 — Transport |
| Connection drops mid-session or won't stay open | Layer 5 — Session |
| "Your connection is not private" / certificate error | Layer 6 — Presentation (TLS) |
| "Server not found," site down, DNS fails, app error | Layer 7 — Application |
How each layer guide helps
This page is the home base. Each guide in the series takes one layer and shows you what breaks there, the exact symptoms, and the commands and fixes to clear it. Work through them in order if you're learning the method, or jump straight to the layer your symptom points to:
- Layer 1 — Physical: cables, ports, NICs, signal and link lights.
- Layer 2 — Data Link: MAC addresses, switches, VLANs and the ARP table.
- Layer 3 — Network: IP addresses, gateways and routing.
- Layer 4 — Transport: TCP, UDP, ports and firewalls.
- Layer 5 — Session: connections that won't stay open.
- Layer 6 — Presentation: TLS and certificate errors.
- Layer 7 — Application: DNS, the browser and the app itself.
📶 Network troubleshooting series — OSI Layers 1–7
🔧 Inspect each layer with our free tools
You can confirm most of these layers in your browser — no install. Match the layer to the tool:
- Layer 2 (Data Link): paste an unknown MAC into the MAC lookup to reveal its manufacturer and identify the device.
- Layer 3 (Network): use the subnet calculator to confirm your IP, mask and gateway line up; the ping test for latency and packet loss; and the IP lookup to see who owns an external IP.
- Layer 6 (Presentation): run the SSL checker to verify a site's certificate is valid and not expired.
- Layer 7 (Application): try Is it DNS? for a one-click DNS health check, the DNS lookup to compare resolvers, and the email check for MX, SPF, DKIM and DMARC.
Layers 1, 4 and 5 are mostly hands-on or CLI work — link lights, port and firewall checks, and session traces.
Stop guessing which layer is broken
Climbing the layers by hand works, but it's slow. Acutis Go runs a 60-second check and tells you plainly which layer the fault is on — the cable, your IP, DNS, or your own machine — so you skip straight to the fix instead of guessing. Free, no account to try.
Get Acutis Go — freeFrequently asked questions
What is the fastest way to troubleshoot a network problem?
Work the OSI model from the bottom up. Start at Layer 1 - the physical cable and link - confirm it is solid, then climb. Each layer depends on the one below it, so a fault at the bottom produces confusing symptoms at the top, and checking in order stops you chasing those.
Why troubleshoot bottom-up instead of starting with the application?
Because a broken lower layer makes every layer above it look broken too. A bad cable can present as a DNS problem, an application timeout, or a slow website - all of which you could spend an hour investigating before discovering nothing was ever going to work over that link.
Do I need to memorise all seven OSI layers?
No. The model is a checklist, not an exam. What matters is the order and roughly what each layer is responsible for, so you can ask the right question at each step and know when to stop climbing.
What is the divide-and-conquer method?
Rather than testing every layer in sequence, you test something in the middle and use the result to eliminate half the stack at once. If a ping to an IP address works, everything from the cable up through routing is healthy and you can concentrate on what sits above it.
Acutis