Network Segmentation for IT Professionals: Your VLAN Table Is a Security Document

·

·

When security consultants assessed Target’s network after the 2013 breach, one finding summed up the whole problem: they could reach cash registers in one store after compromising a deli meat scale in another. Not a server. A scale.

That is what a flat network looks like from an attacker’s side. Once they’re in, everything is one hop away.

If you work in IT, this is good news, because the fix is built from things you already run: VLANs, ACLs, firewall rules, and Group Policy. Network segmentation for IT professionals isn’t a new discipline to learn from scratch. It’s the same infrastructure work you do today, done with a different question in mind. Instead of “how do I connect everything?” the question becomes “what is the minimum connectivity this system needs to do its job?”

This guide covers what segmentation actually is, why it sits at the center of Zero Trust, three changes you can make with the gear you already own, and what mature enterprise segmentation looks like.

Why flat networks keep losing

For decades the model was castle and moat: a hardened perimeter (firewalls, IDS, VPN concentrators) around an internal network where everything trusted everything else.

The problem isn’t the moat. It’s what happens after someone crosses it. The most common way in is still boring: a phishing email, a stolen vendor credential, a reused password. On a flat network, that single foothold inherits reachability to file servers, domain controllers, backup consoles, and payment systems. The attacker doesn’t need a better exploit. They need a network that doesn’t say no.

Two cases every infrastructure person should know:

  • Target (2013). Attackers got in with credentials stolen from an HVAC contractor that used Target’s vendor billing portal. The U.S. Senate Commerce Committee’s staff analysis pointed to weak isolation between the vendor-facing side of the environment and the systems that handled payments as one of the places the attack could have been stopped. Roughly 40 million payment cards were compromised. We break down the full chain in our Breach File on how the Target breach actually happened.
  • NotPetya (2017). The malware spread laterally using the EternalBlue SMBv1 exploit plus stolen credentials, moving across enterprise networks faster than teams could react. A worm like that travels as far as the network lets it. Every segment boundary that blocks SMB is a place it stops. (If EternalBlue is new to you, our WannaCry Breach File explains how it works.)

Neither attack required anything exotic. Both were made far worse by networks where reachability was the default.

What network segmentation is (and why it’s an architecture decision)

Network segmentation is dividing a network into smaller, isolated zones, each with its own access policy governing what traffic can enter and leave.

That definition makes it sound like a configuration task. It isn’t. You can’t segment an environment well without understanding how the business actually works, which means answering three questions first:

  • Data flows: Which applications genuinely need to talk to which databases, and on which ports?
  • User roles: Does marketing really need network-level reachability to the accounting servers?
  • Trust boundaries: What’s the risk tolerance of each environment, and where does one zone’s trust end?

Those answers become the blueprint of your infrastructure. That’s why segmentation is an architectural decision, and why the people best positioned to make it are the ones who already know which undocumented server everything depends on.

Where segmentation fits in Zero Trust

“Zero Trust” has been stretched by marketing until it means almost anything. Strip that away and look at the actual frameworks, NIST SP 800-207 (August 2020) and the CISA Zero Trust Maturity Model 2.0 (April 2023), and one tenet does the heavy lifting: no implicit trust is granted to a user or device based only on where it sits on the network.

Segmentation is how that tenet shows up in the network itself:

  1. Least privilege at the network layer. Users should only reach the files they need. Workloads and devices should only reach the ports and protocols their function requires.
  2. A smaller blast radius. If an HR workstation is compromised, the attacker hits a wall when they try to scan for domain controllers or production databases.
  3. Visibility you didn’t have before. When traffic has to cross a boundary (an internal firewall, or a Layer 3 switch enforcing ACLs), you get a choke point where east-west traffic can be logged, inspected, and alerted on. On a single Layer 2 segment, that traffic is invisible.

We cover the full Zero Trust architecture, including how the policy decision and enforcement roles map to tools you already administer, in Zero Trust for IT Professionals. This post goes one level down: the network work itself.

How a defender thinks through it: Inventory → Threats → Controls → Scale

At Blue Team Academy, we teach a single decision process for defensive work, the Threat & Control Method. Segmentation maps onto it almost step for step, and following the order is what keeps a segmentation project from breaking production.

  • Inventory. What’s actually on the network, what talks to what, and what’s critical? You can’t draw a boundary around flows you haven’t mapped. The output is an Asset Inventory that captures reachability, not just a list of hosts.
  • Threats. If this laptop, this vendor account, or this smart TV is compromised tomorrow, what can it reach? Every path that exists without a business reason is a gift to an attacker. That becomes your Threat Model.
  • Controls. Now choose boundaries and rules against the paths you mapped: VLANs, inter-VLAN firewall policy, host firewalls, management isolation. Ranked, they become your Security Plan.
  • Scale. Apply the same reasoning to the next site, the next cloud account, the next acquisition. That’s what turns one project into a Repeatable Decision Process.

Most failed segmentation projects skip straight to Controls. Keep that in mind as you read the next two sections.

Start here: three wins with the gear you already own

The biggest misconception about segmentation is that it requires expensive micro-segmentation software or a network rebuild. That belief stops a lot of teams from doing anything. You can get real security gains from the managed switches, VLANs, and firewalls already in your racks.

1. Isolate guest Wi-Fi

Visitors, contractors, and personal phones have no business sharing address space with corporate laptops and servers.

The fix: Put guest Wi-Fi on a dedicated VLAN that routes straight to the internet. Block all RFC 1918 (private address) destinations from that subnet so guests can’t even ping internal resources. Turn on client isolation on the wireless controller so guest devices can’t talk to each other either.

2. Corral IoT and building systems

Conference room TVs, cameras, badge readers, printers, and HVAC controllers are known for hardcoded passwords, stale firmware, and insecure protocols. Remember the deli scale.

The fix: Create a dedicated IoT/facilities VLAN. Most of these devices need outbound internet access or a single management server, nothing more. Block them from initiating connections into user or server VLANs. If a printer needs print jobs, allow traffic from the print server to the printer on the printing port it actually uses (commonly TCP 9100 or IPP on 631), and deny the rest.

3. Separate users from servers

On a flat network, a user’s laptop shares a subnet with the file server and the domain controller. If that laptop is infected, it can reach those servers directly at Layer 2, never touching your firewall.

The fix: Put workstations on a Users VLAN and servers on a Servers VLAN, and route between them through your internal firewall. Even if you start with a broad rule while you learn which ports are actually used, forcing the traffic through the firewall gives you logging and IPS inspection you didn’t have yesterday. Then tighten.

Leveling up: enterprise segmentation patterns

Once basic VLAN separation (macro-segmentation) is in place, the next patterns are the ones government and industry guidance keep coming back to, including the NSA’s Network Infrastructure Security Guide (March 2022), the NSA’s 2024 guidance on the network and environment pillar of Zero Trust, and CIS Controls v8.1, which calls for a secure network architecture (Safeguard 12.2) and traffic filtering between network segments (Safeguard 13.4).

1. An isolated management network

Server iLO/iDRAC ports, hypervisor management interfaces, SAN controllers, and switch management IPs are the keys to the kingdom.

The architecture: Never expose management interfaces on production or user VLANs. Put them on a strictly isolated management network. Administrators shouldn’t reach it from their everyday workstation; they should authenticate with MFA to a hardened Privileged Access Workstation (PAW) or a jump host, and reach the management plane only from there.

2. An isolated backup network

Ransomware operators actively hunt for backup servers and NAS devices. Destroying your backups before encrypting production is how they make paying feel like the only option.

The architecture: Give backup infrastructure its own segment. Keep backup servers out of the standard corporate Active Directory domain, so a stolen Domain Admin credential doesn’t also unlock the backup console. Where your backup product supports it, have the backup environment initiate connections to production (pull), rather than letting production systems initiate connections into the backup network. Pair this with a tested restore plan; our ransomware response plan covers the recovery side.

3. Align segments with identity tiers

Microsoft’s legacy Active Directory tier model, since expanded into the broader Enterprise Access Model, groups assets by how much control they hold:

  • Tier 0 (identity and control): domain controllers, PKI, identity providers.
  • Tier 1 (servers and applications): ERP, databases, application servers.
  • Tier 2 (user devices): workstations, laptops, printers.

Network policy should reinforce that separation. Administrative protocols (RDP, WinRM, SMB admin shares) shouldn’t flow from Tier 2 toward Tier 0. That way, a credential stolen from a laptop can’t be used to RDP straight into a domain controller.

4. Micro-segmentation

In heavily virtualized and cloud environments, VLANs get hard to manage at scale. And two web servers in the same VLAN can still attack each other.

The architecture: Micro-segmentation pushes the boundary down to the individual workload, using centrally managed host firewalls or platform-level filtering (VMware NSX, Azure network security groups with application security groups, AWS security groups that reference other security groups). Policy is written against workload identity rather than IP addresses: allow Payroll-Frontend to reach Payroll-DB on TCP 1433; deny frontend-to-frontend traffic.

Pitfalls: how to segment without breaking production

The top reason IT teams avoid segmentation is fear of breaking a legacy application nobody fully understands. That fear is reasonable. Here’s how to avoid a resume-generating event.

  1. Visibility first, enforcement second. That’s the Inventory step, and it’s the one people skip. Don’t start blocking ports blindly. Use NetFlow, firewall logs in log-only mode, or packet captures to baseline traffic for a few weeks, then map dependencies before you write a single DENY rule. Our packet capture analysis workflow is a good companion here.
  2. Beware “any/any.” Moving servers into a new VLAN accomplishes nothing if the rule between VLANs is Allow Any to Any. Every inter-segment rule should name a source, destination, port or protocol, and ideally an application signature.
  3. Don’t forget host firewalls. VLANs stop traffic between segments. They do nothing about lateral movement inside the Users VLAN. Use Group Policy, your MDM, or your EDR platform to enforce Windows Defender Firewall (or the macOS firewall, or nftables/iptables on Linux) and block workstation-to-workstation SMB, RDP, and WinRM. Workstations almost never need to talk to each other.

What this means if you’re moving into security

If you’ve ever built a guest VLAN, written an ACL, or argued about a firewall rule, you’ve already done security architecture work. You just filed it under networking.

That matters, because one of the most common walls IT professionals hit is a resume that still reads like support. “Configured VLANs” sounds like operations. “Designed a segmentation plan that isolated IoT, management, and backup networks and reduced lateral movement paths from user endpoints” sounds like security, and it describes the same work. The skill isn’t new. The framing is.

Cybersecurity is not rocket science. A lot of it is infrastructure you already understand, looked at through the question an attacker would ask. That’s the core of the blue team career path for people who already run infrastructure: not a pile of new tools, but a better reason for using the ones you have.

Start this week

You don’t need a Zero Trust program to begin. Open your VLAN table and ask what each segment can reach that it shouldn’t. Get the conference room TVs off the corporate network. Write down where your most critical data actually lives and which paths lead to it.

Every boundary you add makes an attacker’s job harder, and buys your team time to detect and respond before an incident spreads.

Want the full decision process, not just one application of it? Segmentation is one place the Threat & Control Method shows up. Our program teaches Inventory, Threats, Controls, and Scale step by step, so you can reason through any environment and translate the IT experience you already have into defensive security work.

Not ready for that yet? Get the Keep IT Safe newsletter: practical defensive security breakdowns for people who already run infrastructure. Join Keep IT Safe here.



One response to “Network Segmentation for IT Professionals: Your VLAN Table Is a Security Document”
  1. […] to the database next to it. You still need host-based firewalls or micro-segmentation. Our guide on why your VLAN table is a security document covers that […]