Key takeaways
- First decide: is it “no internet” (the outside world) or “no network” (inside the office)? They have different causes.
- Always work in the same order: physical → IP address → gateway → DNS → ISP.
- Four built-in commands —
ping,ipconfig /all,tracertandnslookup— answer most questions. - If one user is affected, look at that device. If everyone is affected, look at the shared equipment.
“No internet” vs. “no network”
Before touching anything, ask two questions: who is affected, and what exactly doesn't work?
- No network — users can't reach the file server, the printer or the ERP on the local server. The problem is inside the building: a cable, a switch, an IP setting.
- No internet — local resources work, but websites and email don't. The problem is at the edge: the firewall/router, DNS, or the internet provider.
- One user points to that device, its cable or its wall port. One floor or room points to a switch. Everyone points to the core switch, firewall, DHCP/DNS server or the ISP.
The triage order
- Physical. Is the cable plugged in and the port light on? Is the device on Wi-Fi or cable? Did anyone move equipment, unplug a switch or cause a power cut?
- IP address. Run
ipconfig /all. An address beginning with169.254means the device never got an address from DHCP. Check the IP, subnet mask, default gateway and DNS servers look right for your network. - Gateway.
pingthe default gateway (for exampleping 192.168.1.1). If it fails, the problem is between the device and the router — cable, switch, VLAN or Wi-Fi. - Beyond the gateway.
ping 8.8.8.8tests internet reachability without involving DNS. If it works but websites don't load, suspect DNS. - DNS.
nslookup example.comshows whether names are being translated into addresses, and which DNS server answered. - ISP and path.
tracert 8.8.8.8shows where traffic stops. If it dies right after your router, check the modem and contact your provider.
Common problems and what to check first
| Symptom | Likely cause | First check |
|---|---|---|
| Address starts with 169.254 | DHCP server down or unreachable, or DHCP pool full | DHCP service on the server/router; free addresses in the pool |
| “IP address conflict” warning, random disconnections | Two devices with the same address — often a printer or camera set manually inside the DHCP range | Static IPs vs. DHCP range; reserve or exclude static addresses |
| Ping to 8.8.8.8 works, websites don't load | DNS server down or misconfigured | nslookup; DNS settings in ipconfig /all |
| Whole network slow or frozen, switch lights flashing constantly | Network loop / broadcast storm | Recently connected cables, small unmanaged switches, two ports patched together |
| One PC slow or dropping, others fine | Damaged cable, bad wall port, or link at 100 Mbps instead of 1 Gbps | Swap the patch cable and switch port; check link speed |
| Wi-Fi drops, cable is fine | Weak signal, interference, overloaded access point | Signal strength at that desk; number of users on the AP |
| Everything slow at certain hours | Internet line saturated (backups, updates, video) | Bandwidth usage on the firewall |
| Everyone lost internet, local network fine | ISP outage, modem or firewall problem | tracert; modem lights; call the provider |
Slow network and intermittent drops
“Slow” must be made specific: slow to open files on the server, slow websites, or slow everything? Slow file access usually points inside the network — a failing cable, a switch port negotiating at low speed, or the server itself under load. Slow websites with fast local access usually point to the internet line or DNS.
Intermittent problems are the hardest. Run a continuous ping -t to the gateway and to an internet address side by side: if both drop together, the problem is local; if only the internet one drops, it's the line or the provider. Note the times — drops that match a schedule (a backup, a nightly update) are a big clue.
Loops and broadcast storms
A loop happens when two switch ports are connected to each other, directly or through a small desk switch. Traffic circulates endlessly and the whole network can grind to a halt within seconds. Managed switches with spanning tree and loop protection prevent this; cheap unmanaged switches under desks are a common source. Proper design and network segmentation with VLANs also limit how far a storm spreads.
When to call an engineer
- The problem affects everyone and the basic checks don't find it.
- You suspect a loop, a failing switch or a firewall misconfiguration.
- Drops are intermittent and recur over days.
- Nobody has the passwords for the router, firewall or switches — or no one knows how the network is wired.
- You see unknown devices, unusual traffic or other signs of a security problem.
The bottom line
Most network faults are found quickly by following the same order every time: physical, IP, gateway, DNS, ISP. What makes problems drag on is missing documentation, unlabelled cables and equipment nobody can log in to. If your office network is a mystery even to the people who run it, a free IT assessment is a good place to start.
