VLAN Design in Enterprise Networks: Where to Start?

Should VLANs be created by department or by access need? This guide covers the steps of VLAN design, from the access matrix and address plan to trunk settings and a staged move away from a flat network.
Technical summary
Technology
IEEE 802.1Q, VLAN, Layer 3 switch, ACL
Topic
Network segmentation
Level
Beginner–intermediate
Prerequisites
  • Current network topology and IP address list
  • Switch management access
In this article8 sections

In a flat network, all devices are in the same broadcast domain: the printer, file server, IP camera and a guest’s phone talk on the same segment. Malware that infects a client reaches the server directly, a broadcast storm generated by one device affects everyone, and separating which traffic comes from where during a fault takes hours. VLAN design is the first step to solving this. But a good VLAN plan starts on paper, before the switch commands.

1. First write down who needs to reach whom

Plan VLANs not by device type but by access rule. A simple table is enough: groups in the rows (users, servers, management, guests, voice, cameras, printers, production), and in the columns which group they can reach through which service. Groups that share the same rule can stay in the same VLAN; every group with a different rule is a candidate for a separate segment.

  • Users: Internet, business applications, printers
  • Servers: Reachable only on the required ports
  • Management: Switch, AP, firewall and server management interfaces; IT team only
  • Guests: Internet only, no access to the internal network
  • Cameras / IoT: Only to the recording server or cloud service

A mistake you may encounter is skipping this matrix and creating department-based VLANs straight away: an accounting VLAN, a sales VLAN, a production office VLAN. If the three departments have the same access rule, this separation brings no security, only three DHCP scopes, three ACLs and three times the documentation. Department separation makes sense if it really requires a different rule (e.g. only accounting will access the finance application). Otherwise it’s an unnecessary burden.

While writing the matrix, also observe existing traffic. Seeing in firewall logs or switch flow data (NetFlow, sFlow) which device talks to whom shows whether the rule on paper matches reality on site. A forgotten license server or a multifunction device sending scans to a printer becomes the system that opens the first fault ticket on cutover night.

2. Match the address plan to the VLAN number

Linking the VLAN number to the IP subnet saves time during a fault. For example, 10.10.20.0/24 for VLAN 20, 10.10.30.0/24 for VLAN 30 and so on. If there are multiple sites, reserving the second octet for the site (10.site.vlan.0/24) simplifies route summarization and VPN rules in a multi-site setup. Choose the subnet size not by today’s device count but with growth headroom.

Tie the gateway address to the same convention in all VLANs: the first usable address of each subnet (.1) or the last address (.254). Which one doesn’t matter; being consistent does. A technician looking at an IP address on the console should know the VLAN and gateway without doing math. Using a separate range outside the corporate address block for the guest network reduces the chance of a mistyped rule leaking into the internal network. Don’t use ranges common in home networks, such as 192.168.0.0/24 and 192.168.1.0/24, on the corporate side; they collide with a remote user’s home network and access over VPN breaks in odd ways.

3. Numbering and naming standard

Don’t use VLAN 1 for user traffic. On all switches, the same VLAN number should serve the same purpose and carry a meaningful name (e.g. V20-USERS, V90-MGMT). This standard also forms the basis for documentation and handovers.

If VTP is used in Cisco environments, a separate warning is needed. When a switch in VTP server or client mode with a higher revision number is connected to the network, it can push its VLAN database to the entire domain and delete existing VLANs. Losing the VLANs of a whole campus with a switch brought in from the lab is a classic outage scenario in the field. If there are few VLANs and changes are rare, keeping VTP in transparent mode and defining VLANs by hand is more predictable. Check the show vtp status output before connecting a new switch to the network.

4. Trunks, access ports and the native VLAN

  • Endpoint ports should be in access mode, only in the relevant VLAN.
  • Links between switches should be trunks, allowing only the required VLANs (allowed VLAN list).
  • The native VLAN should be moved to an unused VLAN and be the same at both ends.
  • Unused ports should be shut down or put in an empty “parking” VLAN.
  • If IP phones are used, a separate voice VLAN should be defined.

The most expensive single word in trunk configuration is add. In Cisco IOS, to add a new VLAN to an existing trunk you type switchport trunk allowed vlan add 40. If add is forgotten and switchport trunk allowed vlan 40 is typed, the allowed list becomes only VLAN 40 and all other VLANs passing through that trunk are cut immediately. On an uplink trunk, this mistake disconnects everyone attached to that switch. If you’re working remotely, your management session goes too.

A native VLAN mismatch is a quieter problem. If different native VLANs are defined at the two ends of a trunk, untagged traffic lands in another VLAN on the other side; two segments are connected to each other without anyone noticing. CDP logs a warning in this case, so don’t ignore “native VLAN mismatch” lines in switch logs. After a change, verify each trunk’s mode, native VLAN and allowed VLAN list with show interfaces trunk. Before making access decisions between segments, clarify the security boundary through the difference between VLANs and network security.

Turning off DTP on edge ports (switchport mode access and switchport nonegotiate) prevents a user from turning the port into a trunk. BPDU Guard and PortFast should be used together on access ports; when someone connects a small switch at their desk and creates a loop, the port shuts down immediately and the loop doesn’t spread to the whole VLAN.

5. Where will inter-VLAN routing be done?

There are two common models. Routing on a Layer 3 core switch provides high performance; filtering between segments is done with ACLs. Routing on the firewall offers stateful rules, logging and an audit trail; but the firewall’s capacity must be able to carry this traffic. In practice, a common solution is to route heavy traffic such as users and servers on the core, and segments that need control such as guests, cameras and production on the firewall.

Switch ACLs are stateless. If you allow traffic from users to servers and forget the return traffic, the connection isn’t established; if you allow the return broadly, the rule loses its meaning. That’s why ACLs written on the core should be short, readable and commented. Rules for sensitive segments should be kept on the firewall.

The moment you move routing, think about DHCP too. A client’s broadcast-based DHCP request doesn’t cross the VLAN boundary; if DHCP relay (ip helper-address) isn’t defined on the new VLAN’s SVI, devices moved to that VLAN can’t get an address. The first phone call on cutover night usually comes because of this. In the same step, verify that the relevant scope has been created on the DHCP server.

6. Plan the migration without interruption

Moving from a flat network to a VLAN-based structure doesn’t have to happen in one night. Management and guest VLANs are created first, then the new segments; device groups are moved in turn together with their DHCP scopes. Before each step, back up the current configuration and write down the rollback step.

When making changes on remote switches, the rollback plan should be automatic. In Cisco IOS, setting up a scheduled reload with reload in 10 before the change ensures that if you lose access, the device comes back up ten minutes later with the last saved configuration. If the change succeeds, you cancel it with reload cancel. Devices with static IPs (printers, cameras, PLCs, old servers) must be listed before moving; these devices don’t get a new address via DHCP and must be reconfigured by hand. A migration with an incomplete inventory ends with missing devices being found one by one in the morning.

Checklist

  • Access matrix written and approved
  • VLAN–subnet–DHCP scope mapping documented
  • VLAN 1 removed from user traffic, native VLAN changed
  • Allowed VLAN lists defined on trunks
  • Inter-segment rules (ACL/firewall) written on the least privilege principle
  • Configuration backups and topology diagram updated

Frequently asked questions

How many VLANs should we use?

The number isn’t the goal. One VLAN for each group with a different access rule is enough: users, servers, management, guests, voice, cameras, printers and so on. Splitting groups that share the same rule into separate VLANs makes management harder.

Does a VLAN provide security on its own?

No. VLANs separate broadcast domains; if traffic between VLANs is routed, segments can reach each other without ACL or firewall rules. For security, rules must be defined at the routing point.

Why is the native VLAN changed?

The native VLAN, which carries untagged traffic on trunk ports, is VLAN 1 by default. Moving it to an unused VLAN and not using VLAN 1 for user traffic reduces the risk of misconfiguration and VLAN hopping.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Network

We build switching, routing, cabling and wireless networks with VLAN segmentation and redundant uplinks, and provide regular maintenance and fault support.

Knowledge Center

Other articles