Technical analysis

What is DNS? A Beginner's Guide to the Internet's Phone Book

You type google.com and hit enter. Something magically figures out where Google actually lives on the internet — an IP address like 142.250.190.14 — and connects you to it. That something is DNS. Grab a coffee. We're going to explain how the internet's phone book actually works, why it matters in Active Directory environments, and what breaks when it goes wrong.

MUA
MU.A
Microsoft 365 / Entra Engineer
Published
8 min read
Cover image for What is DNS? A Beginner's Guide to the Internet's Phone Book

Grab a coffee. This one's easy to underestimate — DNS is the sort of thing everyone kind of knows about but almost no one truly understands. And in Windows environments, roughly 60% of the "weird problems" you'll ever see are actually DNS problems in disguise.

By the end of this post you'll actually understand how it works. Let's do this.

The problem DNS solves

Computers don't know what "google.com" means. They only understand numbers — specifically, IP addresses like 142.250.190.14.

But humans are terrible at remembering IP addresses. Nobody wants to type 142.250.190.14 to visit Google, 40.113.200.201 for Microsoft, and thousands of other numbers for every website they use.

DNS is the translator. You type a name your brain remembers, DNS turns it into the IP address the network needs.

That's the whole idea. Everything else is details about how.

The phone book analogy (that's actually accurate)

Imagine the pre-internet 1990s. You want to call a pizza place. You don't have their number memorized. So you open the phone book, look up "Pizza Palace," find their number, and dial it.

DNS is exactly that, but for computers.

  • You type pizzapalace.com in your browser
  • Your computer opens the "phone book" (DNS)
  • It looks up pizzapalace.com
  • It gets back an IP address like 203.0.113.42
  • It connects to that IP

Except unlike a physical phone book that lived in one location, DNS is a massive, distributed, hierarchical system running on thousands of servers around the world. And it has to handle billions of lookups per second without falling over.

Vocabulary you need first

Six terms. Learn these and everything else clicks.

Hostname / domain name. The human-readable name — google.com, sentinelidentity.ca, login.microsoftonline.com. What you type.

IP address. The number the network actually uses. IPv4 looks like 142.250.190.14. IPv6 looks like 2607:f8b0:4004:c1b::8a.

Resolver. The thing that does the lookup for you. When you type a URL, your operating system asks a resolver — usually your router at home, or your company's DNS server at work — to find the IP.

Authoritative server. The server that officially knows the answer for a specific domain. sentinelidentity.ca's authoritative servers are wherever the site's owner registered them (usually the hosting provider). Whatever the authoritative server says is the truth.

DNS record. The entry in the phone book. Different types of records store different information. The most common:

  • A record — maps a hostname to an IPv4 address (google.com142.250.190.14)
  • AAAA record — same but for IPv6
  • CNAME — an alias pointing to another hostname (www.example.comexample.com)
  • MX record — mail server for a domain (used by email delivery)
  • TXT record — arbitrary text (used for SPF/DKIM email auth, site verification, and more)

TTL (Time To Live). How long a resolver is allowed to cache the answer before it must ask again. TTLs range from seconds to days.

What actually happens when you type a URL

Let me walk through the full journey when you type sentinelidentity.ca in your browser. This is the moment where DNS actually clicks.

  1. Browser cache check. Your browser looks in its own cache. Did we look this up recently? If yes, use it. Done.
  2. OS cache check. If not, Windows/macOS/Linux checks its own OS-level DNS cache. Same question.
  3. Resolver query. If neither cache has it, your computer asks its configured resolver. At home, that's usually your ISP's DNS server or Google's 8.8.8.8 or Cloudflare's 1.1.1.1. At work, it's usually your company's Domain Controllers (which run DNS).
  4. Recursive lookup. The resolver does the actual hunting on your behalf. It goes:
    • Ask a root server: "who's authoritative for .ca?" Root server responds with the nameservers for Canadian domains.
    • Ask the .ca server: "who's authoritative for sentinelidentity.ca?" .ca server responds with the specific nameservers.
    • Ask the authoritative server: "what's the A record for sentinelidentity.ca?" Authoritative server returns the IP address.
  5. Response cached. The resolver caches the answer (respecting the TTL) and returns it to your computer.
  6. Your computer caches it too. Faster next time.
  7. Browser connects to the IP. Now it can actually load the page.

All of this happens in tens of milliseconds. Usually. When it's slow or broken, everything on your network feels slow or broken — because everything on your network depends on DNS.

Why DNS matters so much in Active Directory

Here's the thing that trips people up: Active Directory requires DNS to function. Not "AD works better with DNS" — AD literally cannot work without DNS. Every domain-joined computer uses DNS to find its Domain Controller. Every Kerberos ticket lookup uses DNS. Every Group Policy download uses DNS.

That's why:

  • The first Domain Controller you install typically becomes a DNS server too. Most AD deployments run DNS on the DCs themselves.
  • Clients in an AD environment must have their DNS pointed at the DC (or another DNS server that can resolve the AD zone). If a computer's DNS points at 8.8.8.8 instead of the DC, that computer can't find its domain and everything breaks.
  • DNS problems look like AD problems. When a user says "I can't access the file share" or "my computer says it can't reach the domain," 60% of the time the underlying cause is DNS misconfiguration.

If you take one thing away from this post: when Windows says "cannot find domain," check DNS first.

What breaks in real deployments

Six DNS problems that cause the vast majority of tickets:

"My computer says it can't reach the domain." DNS server misconfigured on the client. Check ipconfig /all — the DNS server should be your DC, not your ISP.

"I can ping the DC by IP but not by name." Client is using the wrong DNS server, or DNS isn't answering. Try nslookup contoso.com — if it fails, that's your problem.

"External websites are slow but internal stuff is fine." Your internal DNS server is missing a forwarder to a public DNS. When it doesn't have an answer for an external site, it doesn't know where to ask. Add 1.1.1.1 or 8.8.8.8 as a forwarder in your DNS server config.

"I updated a DNS record but nothing sees the change." Caching. Wait for the TTL to expire (or flush client caches with ipconfig /flushdns). If you set the TTL to 24 hours, it'll take 24 hours for everyone to see the change.

"Group Policy isn't applying to some computers." Half the time it's DNS. Domain-joined computers need to resolve _ldap._tcp.dc._msdcs.contoso.com to find a DC that serves policies. Broken DNS breaks GPO application.

"Email isn't being delivered from our company." Your MX or SPF records are wrong. Every email server that receives mail checks these records to decide whether to accept mail from your domain.

Try it in your lab

If you have the home lab from the earlier post, try these:

  1. On CLIENT01, run nslookup dc01.lab.local — should return the DC's IP address.
  2. Change CLIENT01's DNS server to 8.8.8.8 (Google DNS). Then try to open a file share on \\fs01. It won't work. Because Google's DNS doesn't know your internal names.
  3. Point DNS back at the DC. Everything works again.
  4. On the DC, open DNS Manager. Look at the lab.local forward zone — you should see A records for every joined computer.
  5. Delete one of those records. Watch what happens the next time that computer tries to authenticate.

Doing these once cements how DNS underpins AD.

The FAQ

What's the difference between a public DNS server (like 8.8.8.8) and my company's DNS server? Both do DNS. The difference is what they know. Google's DNS knows about internet domains (google.com, github.com, etc.). Your company's DNS knows about your internal domain (contoso.local, and every domain-joined computer's hostname). If you want both, your company's DNS server "forwards" requests it can't answer to a public DNS.

Why is DNS so critical for AD but not (as much) for random web servers? Because AD uses DNS to discover services (DCs, Kerberos, LDAP, Global Catalog) via SRV records. Web servers just have an A record. AD is a much heavier user of DNS than a website is.

Can I run AD without DNS on the DC itself? Technically yes — you can have DNS on a separate server. In practice, everyone runs DNS on the DCs because it simplifies replication and reduces failure points.

What's a "conditional forwarder"? A rule that says "for this specific domain, ask this specific DNS server." Useful when you have multiple AD forests that need to resolve each other's names.

How is DNS different in Entra ID (cloud)? Entra ID doesn't use DNS the same way on-prem AD does. Users authenticate to Entra ID via HTTPS to login.microsoftonline.com, which resolves through public DNS like any website. Internal cloud-only orgs don't need to run their own DNS.

How do I know if DNS is my problem? nslookup your domain name. If it fails or returns wrong answers, DNS is your problem. If it works, look elsewhere.

Where to go next

If DNS makes sense, the natural next posts:

Studying for a cert? DNS is a huge topic in AZ-800 and MS-102. Building a small lab and breaking DNS on purpose is the best way to prep.

Now go grab a coffee. And next time someone at work says "the network is down," check DNS first.

About the author

MUA

MU.AMicrosoft 365 / Entra Engineer

MU.A is a hands-on Microsoft 365 and Microsoft Entra engineer. Sentinel Identity is where MU.A writes the long-form troubleshooting and architecture notes that ship articles people can actually use in production tenants.

Feedback

Was this article helpful?

Anonymous — no account or comment required.

Send detailed feedback

Related reading

How to Promote a Server to a Domain Controller: PowerShell and GUI Walkthrough

You've got a Windows Server ready. It has an IP, a hostname, DNS pointing at the right place. Now you need to actually turn it into a Domain Controller. Grab a coffee. We'll walk through both the PowerShell way (fast, scriptable, professional) and the GUI way (visual, forgiving, good for learning). Then we'll verify it worked and cover what breaks.

How to Build a Physical Domain Controller: A Practical Walkthrough

Every AD deployment guide assumes you already have a running Windows Server. What if you don't? What if you've been handed a physical box and told 'make this a domain controller'? Grab a coffee. We're going from bare metal to fully-functional DC — hardware to Windows install to promotion — in one post.