$title =

Common Mistakes in Configuring Firewalls for Small Businesses

;

$content = [

A small business firewall usually fails because of how it was configured, not because the hardware is weak. A router installed in 2021 with the admin password still set to 'admin', a rule that says 'allow any any', and a remote management page open to the whole internet will let an attacker in no matter how much the device cost. The 2016 Mirai botnet, for example, spread by trying a dictionary of roughly 60 factory-default username and password pairs against internet-facing devices. This guide walks through the most common firewall mistakes and shows how to fix each one, in the order you would work through them on a real network. It follows the guidance in NIST SP 800-41 Rev. 1 (Guidelines on Firewalls and Firewall Policy) and the CIS Critical Security Controls v8, and it uses a fictional 25-person office with the LAN 192.168.1.0/24. The same logic applies to pfSense, OPNsense, Fortinet, SonicWall, Ubiquiti, and most other platforms.

Inventory Your Firewall and Document Current Rules

You cannot fix what you have not looked at. Before changing anything, record the make, model, firmware version, and every rule currently in place. CIS Controls v8 starts with the same step: Control 1 (inventory of enterprise assets) and Control 4 (secure configuration) both assume you know what you are running. Many small businesses inherit a firewall from a former IT contractor and have no idea what half the rules do. Export the configuration and take screenshots of the rule list so you can roll back if a change breaks something.

Build a simple spreadsheet with one row per rule. For each rule, record the source, destination, port, action, who asked for it, and why. Any rule nobody can explain goes into a 'review' pile. If you have a Linux-based firewall, you can dump the active ruleset directly:

# List all active iptables rules with line numbers and packet counters
sudo iptables -L -n -v --line-numbers

# On nftables-based systems
sudo nft list ruleset > firewall-ruleset-2026-11-07.txt

The packet counters are useful. A rule with zero hits after several months is a strong candidate for removal, while a rule with millions of hits deserves a closer look to confirm it is doing what you think.

Change Default Admin Credentials and Enable Multi-Factor Authentication

Default usernames and passwords for nearly every firewall model are published in vendor manuals and public databases, and automated scanners try them first. CIS Controls v8 Safeguard 4.7 specifically calls for managing default accounts on enterprise assets. NIST SP 800-63B favors password length over complexity rules, so use a unique passphrase of at least 16 characters stored in a password manager, and rename or disable the default 'admin' account if the platform allows it. Create individual named accounts for each administrator so that log entries show who did what.

  • Replace the default password with a unique passphrase of 16+ characters, for example four unrelated words plus a number, generated by your password manager.
  • Create a separate named account for each person who administers the firewall, and delete shared accounts.
  • Enable multi-factor authentication (TOTP app or hardware key) on the management interface if the firewall supports it. Current pfSense, OPNsense, FortiGate, and Ubiquiti releases offer TOTP for admin logins, and PCI DSS v4.0 requirement 8.4 expects MFA for administrative access to systems in the cardholder data environment.
  • Set a session timeout of 10 to 15 minutes so an unattended browser tab does not stay logged in.
  • Lock the admin interface to the internal management VLAN or a specific admin workstation IP.

Update Firewall Firmware to the Latest Version

Firewalls are a favorite target because they sit on the edge of the network and a single unpatched flaw can give an attacker direct access. Two examples show how fast this happens. Fortinet disclosed CVE-2022-40684, an authentication bypass in FortiOS, in October 2022 (advisory FG-IR-22-377), and CISA added it to its Known Exploited Vulnerabilities (KEV) catalog the next day because attackers were already using it. In April 2024, Palo Alto Networks disclosed CVE-2024-3400 in PAN-OS GlobalProtect, rated CVSS 10.0, which was exploited as a zero-day before a patch existed. Check the vendor's security advisory page for your model, compare your running version to the latest stable release, and cross-check the CISA KEV catalog for your vendor.

Do the update outside business hours, export a configuration backup first, and read the release notes for changes that affect existing rules. If the device is end-of-life and no longer receives security patches, plan a replacement rather than hoping for the best. On an OPNsense or pfSense system you can check the current version and available updates from the shell:

# OPNsense: check for and apply updates
opnsense-update -c    # check for updates
opnsense-update       # apply updates

# pfSense: show current version
cat /etc/version

Subscribe to the vendor's security mailing list or RSS feed, and to CISA's KEV catalog feed, so you hear about critical patches the day they are released instead of at the next audit.

Replace 'Allow Any' Rules with a Default-Deny Policy

The single most damaging misconfiguration is a rule like 'source any, destination any, port any, allow'. Someone often adds it temporarily to get an application working and never removes it. It makes every other rule irrelevant because traffic matches the permissive rule first.

NIST SP 800-41 Rev. 1 recommends a default-deny stance: block everything, then explicitly allow only what the business needs. PCI DSS v4.0 requirement 1.2.5 likewise requires that every allowed service, protocol, and port be identified, approved, and have a defined business need. Here is what default deny looks like as a minimal iptables policy for a gateway that allows established connections, lets the LAN browse the web and resolve DNS, and drops everything else:

# Set default policies to drop
sudo iptables -P INPUT DROP
sudo iptables -P FORWARD DROP
sudo iptables -P OUTPUT ACCEPT

# Allow loopback and return traffic for connections the firewall already approved
sudo iptables -A INPUT -i lo -j ACCEPT
sudo iptables -A INPUT -m conntrack --ctstate ESTABLISHED,RELATED -j ACCEPT
sudo iptables -A FORWARD -m conntrack --ctstate ESTABLISHED,RELATED -j ACCEPT

# Allow LAN (eth1) to reach the internet (eth0) on web and DNS only
sudo iptables -A FORWARD -i eth1 -o eth0 -p tcp -m multiport --dports 80,443 -j ACCEPT
sudo iptables -A FORWARD -i eth1 -o eth0 -p udp --dport 53 -j ACCEPT
sudo iptables -A FORWARD -i eth1 -o eth0 -p tcp --dport 53 -j ACCEPT

On a GUI-based firewall the idea is the same. Put specific allow rules at the top, and make sure the last rule on each interface is an explicit deny with logging turned on so you can see what is being blocked. Before deleting an old 'allow any' rule, disable it first and watch the logs for a few days to find any legitimate traffic that depended on it.

Close Unused Ports and Disable Unneeded Services

Every open port is a door. CIS Controls v8 Safeguard 4.8 calls for uninstalling or disabling unnecessary services. Over the years, businesses accumulate port forwards for things that no longer exist: an old security camera DVR on port 8080, a retired FTP server on port 21, a remote desktop forward to a PC that was replaced. Go through your NAT and port forwarding table and remove everything that does not map to a current, documented business need. Exposed Remote Desktop Protocol is worth special attention, since the joint CISA and FBI #StopRansomware Guide lists it among the most common initial access vectors for ransomware.

  • Remove port forwards for retired systems, such as an old DVR on TCP 8080 or an FTP server on TCP 21.
  • Never forward RDP (TCP 3389), SMB (TCP 445), or Telnet (TCP 23) directly to the internet. Use a VPN instead.
  • Disable UPnP on the firewall. It lets devices on the LAN open inbound ports automatically without your approval.
  • Turn off services you do not use, such as the built-in DNS forwarder, SNMP v1/v2c (which send community strings in cleartext), or a web server on the firewall itself.
  • If a service must be public, such as a web server, forward only the specific port to a single host in a separate DMZ segment.

Disable Remote Management from the Internet or Restrict It by VPN and IP

Exposing the firewall's own admin page (often HTTPS on port 443, 8443, or 4433) or SSH on port 22 to the internet is a common and serious mistake. CISA considered the problem serious enough to issue Binding Operational Directive 23-02, which requires U.S. federal agencies to remove management interfaces of networked devices from the internet or protect them behind access controls. MITRE ATT&CK tracks the same exposure as technique T1133 (External Remote Services). The safest setting is to disable WAN-side management entirely and manage the device from the inside or over a VPN.

If you truly need remote administration, require a VPN connection first and then allow management only from the VPN subnet. If a VPN is not possible, restrict the rule to the specific public IP of your admin location. For SSH, add key-based authentication and turn off passwords. This sshd_config snippet is a reasonable baseline:

# /etc/ssh/sshd_config
Port 22
PermitRootLogin no
PasswordAuthentication no
PubkeyAuthentication yes
AllowUsers fwadmin
MaxAuthTries 3
LoginGraceTime 30

# Then restart the service
sudo systemctl restart sshd

And a matching firewall rule that permits SSH to the firewall only from the management VLAN (10.10.99.0/24) and drops it from everywhere else:

sudo iptables -A INPUT -p tcp --dport 22 -s 10.10.99.0/24 -j ACCEPT
sudo iptables -A INPUT -p tcp --dport 22 -j DROP

Segment the Network with VLANs for Guests, IoT, and Servers

A flat network, where the front-desk PC, the guest Wi-Fi, the smart thermostat, the security cameras, and the file server all share 192.168.1.0/24, lets an attacker who compromises one device reach all the others. A firewall can only filter traffic that crosses it, so you need separate segments to give it something to enforce. CIS Controls v8 Safeguard 12.2 (secure network architecture) and Safeguard 13.4 (traffic filtering between network segments) describe this approach, and PCI DSS v4.0 requirement 1.4 requires controls between trusted and untrusted networks.

A practical layout for a 25-person office looks like this:

  • VLAN 10, Staff: 192.168.10.0/24. Workstations and laptops. Can reach servers and the internet.
  • VLAN 20, Servers: 192.168.20.0/24. File server and accounting system. Accepts connections only from VLAN 10 on required ports, such as TCP 445 to the file server.
  • VLAN 30, Guest: 192.168.30.0/24. Internet only. Blocked from all internal ranges.
  • VLAN 40, IoT and cameras: 192.168.40.0/24. No access to staff or servers. Allowed to reach only the camera recorder and required cloud endpoints.
  • VLAN 99, Management: 10.10.99.0/24. Firewall, switch, and access point admin interfaces only.

Then write the inter-VLAN rules. This example blocks the guest VLAN from every private range while still allowing internet access:

# Block guest VLAN (192.168.30.0/24) from reaching any private address space
sudo iptables -A FORWARD -s 192.168.30.0/24 -d 10.0.0.0/8 -j DROP
sudo iptables -A FORWARD -s 192.168.30.0/24 -d 172.16.0.0/12 -j DROP
sudo iptables -A FORWARD -s 192.168.30.0/24 -d 192.168.0.0/16 -j DROP

# Allow guest VLAN out to the internet
sudo iptables -A FORWARD -s 192.168.30.0/24 -o eth0 -j ACCEPT

Your managed switch and access points must support IEEE 802.1Q VLAN tagging, and each Wi-Fi SSID should map to the correct VLAN. Confirm this by connecting a phone to the guest network and verifying it cannot ping or open a file share on VLAN 20.

Restrict Outbound Traffic to Only Required Destinations

Most small businesses filter inbound traffic and leave outbound wide open. That is a gap, because malware that gets in needs to call home to a command-and-control server, and data theft needs a path out. MITRE ATT&CK catalogs these behaviors as Command and Control (TA0011) and Exfiltration (TA0010), and egress filtering shrinks that opportunity.

Start by restricting what each segment may send out. Servers rarely need general web browsing. IoT devices usually need a handful of vendor domains. Staff workstations need web, email, and DNS. Also force all DNS through your own resolver so malware cannot bypass filtering by using a hard-coded external one:

# Servers VLAN: allow only NTP, DNS to internal resolver, and HTTPS to update servers
sudo iptables -A FORWARD -s 192.168.20.0/24 -p udp --dport 123 -j ACCEPT
sudo iptables -A FORWARD -s 192.168.20.0/24 -d 192.168.10.2 -p udp --dport 53 -j ACCEPT
sudo iptables -A FORWARD -s 192.168.20.0/24 -p tcp --dport 443 -j ACCEPT
sudo iptables -A FORWARD -s 192.168.20.0/24 -j DROP

# Staff VLAN: redirect any outbound DNS to the internal resolver at 192.168.10.2
sudo iptables -t nat -A PREROUTING -s 192.168.10.0/24 ! -d 192.168.10.2 -p udp --dport 53 -j DNAT --to-destination 192.168.10.2

# Block common outbound ports used by malware and legacy file sharing
sudo iptables -A FORWARD -s 192.168.10.0/24 -p tcp -m multiport --dports 23,135,139,445 -o eth0 -j DROP

Roll this out gradually. Begin in log-only mode, review a week of traffic, and then switch to blocking so you do not break a payroll provider or software license check on a Monday morning.

Enable Logging and Send Alerts for Suspicious Activity

A firewall that does not log gives you no way to know it has been probed or breached. At a minimum, log denied traffic, administrative logins (successes and failures), and configuration changes. CIS Controls v8 Safeguard 8.10 sets a 90-day minimum for audit log retention, and Safeguard 8.9 calls for centralizing logs. If you handle card payments, PCI DSS v4.0 requirement 10.5.1 asks for 12 months of retention with the most recent 3 months immediately available. Store logs off the device so an attacker who compromises the firewall cannot erase the evidence.

Forward logs to a syslog server or a cloud log service. On a Linux gateway, this rsyslog line sends everything to a collector at 192.168.20.50, and the iptables rule logs dropped inbound packets with a recognizable prefix:

# /etc/rsyslog.d/50-forward.conf
*.* @@192.168.20.50:514

# Log dropped inbound packets (rate-limited to avoid flooding)
sudo iptables -A INPUT -m limit --limit 5/min -j LOG --log-prefix "FW-DROP-IN: " --log-level 4

# Restart rsyslog
sudo systemctl restart rsyslog

Set up alerts for events that matter: more than 5 failed admin logins in 10 minutes, a new firewall rule created, a configuration change outside business hours, or a single external IP hitting dozens of closed ports. Email or chat alerts to a monitored mailbox are enough for a small team, as long as someone actually reads them.

Test Rules with a Port Scan and Verify Nothing Unexpected Is Exposed

After any change, verify the result instead of assuming the rules work. The most reliable check is a scan from outside your network, for example from a phone hotspot or a cloud VM, against your public IP. NIST SP 800-115 (Technical Guide to Information Security Testing and Assessment) covers this kind of network port and service identification. Only scan addresses you own or are authorized to test.

# Scan the 1000 most common TCP ports on your public IP (203.0.113.10 is an example)
nmap -Pn -sS --top-ports 1000 203.0.113.10

# Full TCP port sweep (slower, run off-hours)
nmap -Pn -sS -p- -T4 203.0.113.10

# Check common UDP services
sudo nmap -Pn -sU --top-ports 50 203.0.113.10

A healthy result for an office with no public servers looks like this:

Starting Nmap 7.94 ( https://nmap.org )
Nmap scan report for 203.0.113.10
Host is up (0.021s latency).
All 1000 scanned ports on 203.0.113.10 are in ignored states.
Not shown: 1000 filtered tcp ports (no-response)

Nmap done: 1 IP address (1 host up) scanned in 21.43 seconds

If the scan instead shows something like '3389/tcp open ms-wbt-server' or '8443/tcp open https-alt', you have found an exposure to fix right away. Also test from the inside: confirm a guest device cannot reach the server VLAN, that blocked outbound ports really fail, and that a legitimate workflow, such as accounting software or VoIP, still works after your changes.

Back Up the Configuration and Schedule Regular Rule Reviews

Once the firewall is in good shape, protect that work. Export the configuration after every approved change, name the file with the date, and store copies somewhere other than the firewall, such as an encrypted folder on your backup system. If the device fails or a bad change locks you out, a recent backup turns a day of rebuilding into a 20-minute restore.

Then put recurring reviews on the calendar so the configuration does not drift back into the state you started with. PCI DSS v4.0 requirement 1.2.7 sets a useful baseline by requiring network security control configurations to be reviewed at least every six months:

  • Monthly: check for firmware updates and security advisories, compare against the CISA KEV catalog, and skim the denied-traffic logs for patterns.
  • Quarterly: walk through every rule in your spreadsheet, remove anything without a current owner or purpose, and check hit counters for rules that never match.
  • Twice a year: run an external port scan and compare the results to your documented list of intended exposures, which also satisfies the six-month rule review cadence.
  • After every staff change: remove accounts and VPN access for anyone who has left, and rotate shared secrets they knew.
  • After every new system or vendor: add the rule with a written reason and an expiry or review date.

Treat each change as a small ticket: who requested it, what was changed, when, and how it was tested. That habit alone prevents most of the problems described above, because rules stop being mysterious and temporary exceptions stop becoming permanent holes.

References and Further Reading

  • NIST SP 800-41 Rev. 1, Guidelines on Firewalls and Firewall Policy (csrc.nist.gov).
  • NIST SP 800-63B, Digital Identity Guidelines: Authentication and Lifecycle Management (csrc.nist.gov).
  • NIST SP 800-115, Technical Guide to Information Security Testing and Assessment (csrc.nist.gov).
  • CIS Critical Security Controls v8, Controls 1, 4, 8, 12, and 13 (cisecurity.org).
  • PCI DSS v4.0, Requirements 1, 8, and 10 (pcisecuritystandards.org).
  • CISA Binding Operational Directive 23-02, Mitigating the Risk from Internet-Exposed Management Interfaces (cisa.gov).
  • CISA Known Exploited Vulnerabilities Catalog, including CVE-2022-40684 (Fortinet FortiOS) and CVE-2024-3400 (Palo Alto PAN-OS) (cisa.gov).
  • CISA and FBI #StopRansomware Guide (cisa.gov/stopransomware).
  • MITRE ATT&CK techniques T1133 (External Remote Services), TA0010 (Exfiltration), and TA0011 (Command and Control) (attack.mitre.org).

];

$date =

;

$category =

;

$author =

;