Decisions made in the first week of a new FortiGate determine how the device will be managed for years. Rules written against interfaces instead of zones, a management port left open to the internet, or a NAT mode that’s hard to change later don’t bother anyone on day one; two years later they make every change risky. The checks below collect what to look at before and after installation under management access, zone design, NAT, security profiles, VPN, logging and lifecycle. The same principles can be applied, with their own terminology, to other platforms such as Palo Alto Networks, Cisco, Check Point or Sophos.
1. Lock down the management plane from the start
The firewall itself is one of the most valuable targets on the network it’s meant to protect. Firewall vendors’ security advisories regularly include vulnerabilities affecting the management interface and the VPN portal; a management interface reachable from the internet is a direct target for these vulnerabilities. That’s why management access should be restricted in the first step of the installation.
- Disable HTTPS and SSH management access on WAN interfaces; allow access only from the management network or specific management workstations (trusted host definitions in FortiOS).
- Replace the default
adminaccount with personal administrator accounts; each administrator should have their own account and MFA. - If SNMP is used, prefer SNMPv3 and restrict community strings and allowed sources.
- Disable unused management services (HTTP, Telnet).
In the FortiOS CLI, which interface responds to which management protocol is determined by each interface’s allowaccess value under config system interface. On WAN interfaces this value shouldn’t include https, ssh or http; if needed for monitoring, only ping can remain. The trusted host restriction is defined per administrator under config system admin with fields such as trustedhost1.
A common mistake is this: trusted hosts are defined for one administrator and left empty on the other accounts. An account left empty accepts logins from anywhere. The restriction must be applied to all administrator accounts, including the local account used for emergencies.
2. Design zones before policy
Instead of using interfaces directly in policies, defining zones for LAN, servers, DMZ, guests, management and, if needed, production (OT) makes the rule base readable and extensible. Which traffic is needed between each pair of zones should be set out in a table before any rules are written.
| Source zone | Destination zone | Example permission |
|---|---|---|
| LAN | Server | Application ports only (e.g. HTTPS, SMB; RDP only for the management group) |
| Guest | WAN | Internet; no access to the corporate network |
| Server | WAN | Only updates and defined external services |
| Management | All | Management protocols, only from management workstations |
The practical benefit of using zones shows up when a new VLAN is added. The VLAN interface is added to the existing “LAN” zone and the existing rules cover it too; an interface-based setup requires every rule to be copied. Tie rule naming, address objects and service groups to a standard from day one too. Finding the right rule among hundreds of “Policy 47”s takes longer than the fault itself.
3. NAT and exposed services
Every service exposed to the internet through port forwarding (VIP / destination NAT) is an attack surface. For each VIP, the business owner, the justification and, if possible, a source IP restriction should be recorded. Instead of exposing management protocols such as Remote Desktop (RDP) directly to the internet, VPN or ZTNA should be used.
FortiOS offers two NAT models: policy-based NAT (the default) and central NAT. This choice should be made at the start of the installation; changing the model after the rule base has grown means rebuilding the entire NAT setup. A VIP object doesn’t open traffic on its own; it needs an allow rule that uses it. Instead of leaving the rule’s source as “all”, if the service is open only to specific business partners, restrict it to their addresses. Unused VIPs pile up over time. Each one is an open door.
4. Attach security profiles to every allow rule
An allow rule only lets traffic through; it’s the IPS, application control, web filtering, DNS filtering and antivirus profiles that inspect the content. Not applying these profiles to internet egress rules is a common gap. Seeing most encrypted traffic requires SSL inspection (deep inspection), which in turn requires planning certificate distribution and an exception list (e.g. banking and healthcare sites). In capacity planning, take into account that the device’s real capacity drops with IPS and SSL inspection enabled.
The built-in certificate-inspection profile in FortiOS only looks at certificate and SNI information; content isn’t visible. The deep-inspection profile decrypts traffic, but clients need to trust the FortiGate’s CA certificate. If deep inspection is enabled before this certificate is distributed through Active Directory GPO or MDM, users get certificate warnings and the help desk is swamped one morning. Start with a pilot group first and mature the exception list there.
5. VPN: MFA and IPsec
MFA should be mandatory for remote access, and only the required destinations should be allowed per user group. With FortiOS 7.6.3, Fortinet removed SSL VPN tunnel mode and recommends IPsec VPN; starting new installations with IKEv2-based IPsec removes the need to migrate in later upgrades. For details, see our VPN and secure remote access page.
Leaving a VPN user with an “all” destination into the internal network turns the firewall into a door into the corporate network. Define destinations per user group: accounting to the ERP server, the software team to the development environment. The split tunnel decision should also be made consciously; with a full tunnel all internet traffic passes through headquarters and consumes line capacity, while with a split tunnel the endpoint’s internet traffic doesn’t pass through the corporate security profiles.
6. Logging and monitoring
- Enable logging on allow and deny rules; at a minimum, internet-facing services, VPN and management access should be logged.
- Send logs not to the device itself but to a central log system (FortiAnalyzer, syslog or a SIEM); time synchronization (NTP) must be correct.
- Define alarms for failed management logins, configuration changes and HA state changes.
If the clock is wrong, logs aren’t evidence. In an incident investigation, when you put firewall, domain controller and server logs side by side, even a few minutes of drift breaks the sequence of events. Calculate log volume from the start as well: with session logging on for all allow rules, the daily data volume can fill a small log server’s disk quickly. Set the retention period according to your internal policy and the regulations you’re subject to.
7. Lifecycle: firmware, backup and HA
Choose the firmware version from the stable versions recommended by the vendor, follow security advisories and upgrade with a test plan. Store configuration backups automatically and off the device. If high availability (HA) is used, regularly verify the synchronization state and the failover test.
Fortinet provides an upgrade path tool that publishes the intermediate versions to follow when jumping between versions; if several major versions are skipped at once, part of the configuration may not be converted. Back up the configuration before upgrading and read the “known issues” section in the release notes. In an HA cluster, get system ha status shows the cluster members and synchronization state, and diagnose sys ha checksum cluster shows whether the configuration checksums of the two devices match. Don’t leave the heartbeat link on a single cable; when a single heartbeat drops, both devices may think they’re primary. That’s split-brain.
Post-installation checklist
- Is the management interface unreachable from the WAN side?
- Does every administrator have a personal account and MFA?
- Are any allow rules with “all/any” services left?
- Are IPS, application and web filtering profiles applied to internet egress rules?
- Is a business owner and justification recorded for every exposed VIP?
- Are logs reaching the central system, and is the time correct?
- Is the configuration backup stored off the device, and has restoring it been tested?
To review an existing firewall’s rule base or plan a new installation, see our Firewall Installation and Rule Management service; for internal network separation, we recommend planning it together with network segmentation.
Show the physical network counterpart of the access zones on the firewall in the enterprise VLAN setup and network segmentation plan.