Zones first: a firewall policy you can still read in three years

By the NexusSec team · 6 min read · Published October 2026
Short version: Decide the zones before writing a single rule: group interfaces and VLANs by trust and purpose (staff, servers, DMZ, guest, IoT, management, VPN users), deny everything between zones by default, and write a rule for each direction that names the application and the group allowed. A rule base built this way stays short enough to review, and a compromised device stays inside its zone.

Ask to see a firewall rule base that has been in production for three years and you will usually find a long list, a good share of it some form of “any to any” on a port, and nobody who can say what most of the rules are for. That is not carelessness. It is what happens when rules are added one ticket at a time with no structure to hang them on. Zones are that structure.

What a zone is

A zone is a group of interfaces, VLANs or VPN tunnels that share a level of trust and a purpose. The firewall writes policy between zones rather than between individual addresses: traffic that stays inside a zone is left to the switches, and traffic that crosses a zone boundary is where the firewall decides.

Most next-generation firewalls are built around zones under one name or another. Firewalls that write rules per interface, such as pfSense and OPNsense, can be planned the same way by grouping interfaces. The vendor matters less than the plan.

A starting set of zones

ZoneWhat goes in itWhat it should reach
StaffEmployees' desktops and laptopsThe internet through inspection; named applications in the server zone; nothing in management
ServersFile server, ERP, domain controllers, internal applicationsUpdates, backups and named partners. Rarely anything in the staff zone
DMZAnything reachable from the internet: web server, mail relay, reverse proxyInbound on published ports only; into internal zones only by named exception
GuestVisitor Wi-FiThe internet, rate-limited. Nothing internal
IoTCCTV recorders, attendance terminals, printers, building systemsTheir own recorder or vendor cloud, and nothing else
ManagementAdmin interfaces of the firewall, switches, hypervisors and storageReached only from administrators' machines or a jump host
VPN usersRemote staff on the remote-access VPNThe same as the staff zone or narrower, never broader

A small office does not need all seven on day one. Guest, IoT and management give the most for the least effort, because they hold the devices nobody looks after and the interfaces most worth protecting. We go through the device side of that in guest Wi-Fi, cameras and printers.

Default deny between zones, one rule per direction

The rule that does most of the work is the last one: anything not explicitly allowed between two zones is dropped and logged. Every allowed flow then gets its own rule, in its own direction. Staff to Servers for the ERP is one rule. Servers to Staff is a separate question, and the answer is usually no. Writing direction explicitly is what stops a compromised server from becoming access to every desktop.

Where the firewall can identify applications and users, write rules for those rather than for ports and IP addresses: “Accounts team, Staff to Servers, ERP” says what it is for, and keeps working when an address changes.

Naming that explains itself

Keeping it clean

Most firewalls can show how often each rule has matched and when it last did. Review that every quarter: rules with no hits for months are candidates for removal, and a rule hidden beneath a broader one above it never matches at all. Our firewall audit and rule review is this exercise done independently.

When the business changes, with a new SaaS tool, a branch or a vendor that needs remote access, add the flow to the zone pair it belongs to rather than to whichever rule sits nearest the top.

Moving an existing firewall to zones

You do not need to rebuild the policy to get most of the benefit. Three changes, in this order, do the bulk of it:

  1. Move guest Wi-Fi and IoT devices into their own zones with internet-only or device-specific rules.
  2. Put management interfaces in a management zone reachable only from administrators' machines.
  3. Between staff and servers, write rules for the flows you know about, put a logging allow rule beneath them for a week to see what else crosses, then turn that last rule into a deny.

If the network is flat today, one laptop should not reach everything covers why this matters, and network segmentation is how we phase it so nothing breaks.

A note on NXSgate

NXSgate, the firewall NexusSec is building, is designed around exactly this model: zones, a default deny between them, and a rule for each direction. It has not been released yet. Everything above applies to whichever firewall you run today.

Frequently asked questions

How many zones should a small office have?

Start with four: staff, servers, guest and management. Add an IoT zone as soon as there are cameras, attendance terminals or networked printers, which in most offices is immediately. More zones are worth it only if someone will maintain the rules between them.

Is a VLAN the same as a firewall zone?

No. A VLAN separates traffic at layer 2; a zone is where the firewall applies policy. Putting each VLAN in a zone, and routing between VLANs through the firewall rather than the core switch, is what makes the separation enforceable.

Should the firewall inspect traffic inside a zone?

Usually not. If devices in the same zone should not talk to each other, as on guest Wi-Fi, use client isolation on the access points or split the zone in two.

Can we move to zones without downtime?

Yes, if it is phased. Create the zones and rules first, move one group of devices at a time in a maintenance window, and keep a logging catch-all rule until the logs show nothing unexpected.

Want your rule base reviewed against a zone plan?

We review the current policy, propose zones that fit how your business works, and phase the change so nothing breaks.

Book a free consultation