NewAI Content Creation is now live in Early Access
Turning Point Training Instituteby Training Center
Skip to content
0%
Module 9 — Network TroubleshootingLesson 3 of 4
Video lesson

DNS Problems and Slow Networks

BeginnerVideo lesson
10 min readBeginner

Where you are: Module 9 — Network Troubleshooting · Lesson 9.3 · Based on Chapter 9 (§9.4–9.6) of Dr. Tahseen Al Doori's Troubleshooting Essentials.

What you'll be able to do

  • Prove that a fault is caused by name resolution (DNS) using the five-command connectivity test.
  • Use nslookup, flush the DNS cache and check the configured DNS servers.
  • Use a temporary public DNS server as an isolating test, within organisational policy.
  • Investigate a slow network in a logical order and match common symptoms to their most likely layer.

Watch: DNS Problems and Slow Networks

The video follows a fictional reception PC that cannot open any website, runs the five-command connectivity sequence from Module 7 along the path to the gateway, the internet and the DNS server, and proves a DNS fault with nslookup. It then uses tracert to locate a delay in a slow network and closes with the book's symptoms table. As you watch, consider:

  • Which command in the sequence fails first, and what does that prove about the layers below it?
  • Why is querying a public DNS server a test rather than a fix?
  • In the tracert output, why does it matter that the delay continues to the destination?

What DNS does

DNS, the Domain Name System, translates names such as example.com into the IP addresses that computers actually use to communicate. Every time a user types a web address, opens email or connects to a cloud service, a DNS lookup usually happens first. If it fails, the user sees a browser message such as "this site can't be reached", even though the network itself may be perfectly healthy.

DNS is an application-layer protocol. That is why the book's symptoms table places "websites by name fail, by IP work" at the Application (DNS) layer, even though, to users, DNS feels like part of the network plumbing.

The signature of a DNS fault

The book gives one clear test:

If websites can be reached by IP address but not by name, the fault is in name resolution.

The five-command connectivity sequence from Module 7 proves this step by step:

code
ping 127.0.0.1           # 1. Is the TCP/IP stack working locally?
ipconfig                 # 2. Do I have a valid IP address and gateway?
ping <default gateway>   # 3. Can I reach my router?
ping 8.8.8.8             # 4. Can I reach the internet by IP address?
ping google.com          # 5. Does name resolution (DNS) work?

(8.8.8.8 is Google's public DNS service, used by the book as a well-known address on the internet.) If steps 1 to 4 succeed and step 5 fails, everything works except turning names into addresses. The first failure shows the layer at fault: this is the half-split method applied to the network path.

The book's DNS checks

Test with nslookupFlush the DNS cacheCheck the DNS serversTemporary public DNS testCorrect and verify
  1. Test with nslookup. It asks the configured DNS server for a name and shows which server it asked. A timeout, or an error, points to that server.
  2. Flush the DNS cache with ipconfig /flushdns on Windows. Computers remember recent answers, including failed ones; flushing throws them away so the next lookup is fresh.
  3. Check the configured DNS servers with ipconfig /all. A server address typed in by mistake, or one that no longer exists, shows up here.
  4. Temporarily set a public DNS server as a test. If names suddenly resolve, the fault is isolated to the original DNS service.
  5. Correct the real setting and verify with the user.

The book's appendix gives the cache-flush commands for the other platforms: sudo dscacheutil -flushcache on macOS and resolvectl flush-caches on Linux.

Using a public DNS server responsibly

A public resolver is a diagnostic tool, not the answer. Three cautions:

  • It is a test, not the fix. It shows where the fault is; the fix is restoring the correct DNS setting or repairing the DNS service.
  • Follow policy. In an organisation, the company's own DNS servers may be the only ones that know internal names, such as file servers or intranet sites, and policy may require them. Changing DNS on a managed device can also be beyond your authority; if so, escalate.
  • Undo and record. Put the correct setting back, verify, and document what you changed and why.

Beyond the book (2026 note): You can often run the isolating test without changing any settings. In non-interactive mode, nslookup takes the name to look up and, optionally, a second argument naming the DNS server to ask, for example nslookup www.example.com 8.8.8.8 (nslookup, Microsoft Learn). If that server answers while the configured one times out, the fault is isolated to the configured DNS service. In PowerShell on Windows, Resolve-DnsName -Name www.example.com -Server 8.8.8.8 does the same job (Resolve-DnsName), and Clear-DnsClientCache is documented as equivalent to ipconfig /flushdns (Clear-DnsClientCache). On Linux systems that use systemd-resolved, resolvectl status shows the DNS servers currently in effect and resolvectl query <name> performs a lookup (resolvectl(1), man7.org).

Slow networks

Not every network fault is a total failure. When the network works but slowly, the book gives five checks:

  1. Test speed on a wired connection. This separates Wi-Fi performance from internet performance. If wired is fast and Wi-Fi is slow, look at the wireless checks from Lesson 9.2.
  2. Look for bandwidth users: large downloads, backups or streaming on any device sharing the connection.
  3. Use tracert (traceroute on Linux and macOS) to find hops with large delays. A hop is one router along the path.
  4. Restart network equipment and check for firmware updates.
  5. Compare measured speeds with the service plan, and contact the provider if they fall short, with your measurements ready.

When reading tracert, look for the hop where the times jump and stay high all the way to the destination. That is where the delay begins. A jump at hop 1 or 2 points inside your own network; a jump further out points to the provider or beyond.

Common symptoms and their likely layers

The book's summary table is worth keeping beside you:

SymptomMost likely layerFirst checks
No connection at all, no link lightPhysicalCable, port, power to switch or router
Connected to Wi-Fi but no internetNetworkIP address, gateway, router and modem
Websites by name fail, by IP workApplication (DNS)DNS settings, flush cache, nslookup
One application cannot connectTransport/ApplicationFirewall rules, ports, application settings
Connection drops intermittentlyPhysical/Data LinkSignal strength, interference, faulty cable

Worked example

(Fictional scenario.) The reception PC cannot open any website; the browser says the site cannot be reached. Everyone else in the building is working. Yesterday a contractor typed static network settings into this PC by hand.

  1. Identify. One device; recent manual change; the symptom is at layer 7.
  2. Test with the five commands. ping 127.0.0.1 succeeds. ipconfig shows 192.168.1.37 with gateway 192.168.1.1. The gateway replies. ping 8.8.8.8 replies. ping www.example.com fails with Ping request could not find host. Only name resolution has failed.
  3. Narrow it. nslookup www.example.com times out and shows that it asked 192.0.2.53, the address the contractor typed in. That is not the office's DNS service.
  4. Isolate. After ipconfig /flushdns, the same lookup sent directly to a public server answers at once. The fault is the configured DNS server, not the internet or the PC.
  5. Plan and implement. With the office manager's approval, the PC's addressing is set back to automatic (DHCP), so it receives the correct DNS server from the router.
  6. Verify. Names resolve; the receptionist opens the booking system and email.
  7. Document. The ticket records the cause, the correction and a request that contractors use the standard configuration.

Try it

Safe, read-only activity on your own computer or a lab machine:

  1. Run the five-command connectivity sequence and record the result of each step.
  2. Run nslookup for a well-known website and note which DNS server answered.
  3. Run the same lookup against a named public resolver (for example nslookup example.com 8.8.8.8) and compare the answers.
  4. Run tracert (Windows) or traceroute (macOS, Linux) to the same site, and identify the first hop outside your own network.

You have finished when you have a short table of the five steps with pass or fail, the name or address of your DNS server, and one sentence describing where your tracert output leaves your local network.

Knowledge check

ping 8.8.8.8 replies, but ping www.example.com cannot find the host. Which layer is at fault?

Why is setting a public DNS server temporarily described as a test rather than a fix?

Wi-Fi feels slow. A speed test over a cable is fast. What has this shown?

In a tracert, hops 1 and 2 respond in under 10 ms, hop 3 jumps to 180 ms and the delay continues to the destination. Where does the delay begin?

Answers explained

  1. c — replies by IP address prove the cable, link and gateway work; only name resolution, an application-layer DNS function, is failing.
  2. a — the public resolver isolates the fault to the original DNS service; public servers do answer correctly, nothing is disabled, and it works on any connection type, but the real setting must still be corrected within policy.
  3. b — the book's first slow-network check separates Wi-Fi from internet performance; a fast wired result clears the provider and DNS, and nothing points to the operating system.
  4. d — the first hop where times jump and stay high marks where the delay starts; hops 1 and 2, including the office router, are fast.

Key takeaways

  • Reach it by IP but not by name: the fault is DNS, an application-layer service.
  • Test with nslookup, flush the cache and check which DNS servers are configured.
  • A temporary public DNS server isolates the fault; it is a test, not the fix, and policy comes first.
  • For slow networks: test wired, find bandwidth users, use tracert, restart and update, compare with the plan.

Sources