This sheet is past its review date. Deadlines or advice may have changed since 5 Nov 2026.
Edge devices: would you know if your firewall or VPN had been broken into?
This month, take every edge device's management interface off the internet and send its logs to a system the device can't erase, and replace anything past end of support within 12 months. When a flaw in one is exploited, assume the device is compromised: collect evidence, reset or rebuild it and rotate every credential it held, because a patch doesn't remove an attacker who is already in.
If you only do five things
Know every device that faces the internet.
List each firewall, VPN gateway, router, load balancer and file-transfer appliance with its software version and end-of-support date, and scan your address ranges for the ones nobody listed. CISA, the FBI and the NCSC urge every organization to keep that inventory1.
Send the logs somewhere the device can't erase.
Forward detailed logs to a central system and alert on admin logins, configuration changes and hardware changes3. Keep them for at least 12 months, as Mandiant advises for network devices' admin logs: BRICKSTORM intrusions averaged 393 days of dwell time6.
Replace devices before support ends.
An unsupported device won't get the fix for the next exploited flaw. Budget each replacement a year before its end-of-support date, and add compensating controls for any you can't remove in time3.
Check a device after an exploited flaw
Collect evidence first, patch within the patching sheet's deadlines, then check every device the flaw could reach before the patch.
Capture evidence before you patch or reboot.
Before updating, store logs off the device and take a disk and memory snapshot, as CISA required for Cisco SD-WAN8, and export the configuration. If you find signs of compromise, disconnect the device but keep it powered on9.
Don't trust a clean integrity check.
CISA warns that Ivanti's checker can miss attacker commands that run only in memory11. On FortiGates, a file link planted by attackers could survive patching12.
Look for new admin accounts and changed settings.
In attacks Fortinet disclosed in January 2026, attackers downloaded configurations and added local admins such as
auditandbackup13. Compare configurations and firmware hashes with your offline copies and the vendor's values15.Reset or rebuild, then upgrade.
Follow the vendor's recovery steps, such as a factory reset or new virtual instance for Ivanti11 or a reimage for Cisco Firepower10. Review any configuration you restore.
Rotate every secret the device held.
Reset the domain, admin and local account passwords, reissue stored certificates and replace API keys and license server credentials11. End live sessions too, as Citrix's
killcommands did after CVE-2025-577714.Hunt behind the device.
Check what the device's accounts and VPN users did in directory, server and cloud logs since the first known exploitation. Attackers can use access they set up months or years later4.
Harden what stays exposed
- Switch off what you don't use. Disable unused ports, protocols and features, including management protocols154. FortiGates that had never enabled SSL-VPN weren't affected by the file-link persistence12.
- Use only encrypted management. Allow SSH, HTTPS and SNMPv3, and block outbound connections from the management interface except to your AAA, syslog and telemetry collectors15.
- Turn off vendor cloud logins you don't need. A FortiCloud single sign-on bypass let attackers into FortiGates that had it enabled13. If you keep such a feature, log and review every use.
- Give the device as few secrets as possible. Bind it to the directory with a dedicated low-privilege account over LDAPS. Attackers use the devices' own packet capture to collect plaintext credentials6.
- Keep it away from the core. Zones, firewall rules and admin paths are in the network segmentation sheet.
What to demand from vendors
- An end-of-support date before you buy. From 11 December 2027, the EU Cyber Resilience Act requires manufacturers of new products to state it at the time of purchase17. Put it in the contract.
- Logs and evidence built in. Ask for TLS-protected remote logging, a record of admin actions and a supported way to collect memory and disk, as the NCSC specifies for network devices16.
- A way to prove the device is clean. Ask for an integrity tool and a recovery procedure that removes persistence. On Cisco Firepower, neither a reboot nor the patch did9.
- Secure management out of the box. Phishing-resistant MFA for admins and no management interface on the internet by default. Canada's Cyber Centre says customers should demand that vendors harden devices to secure management interfaces3.
What changed in 2026
- CISA set deadlines for end-of-support edge devices. BOD 26-02 of February 2026 gives US federal civilian agencies 12 to 18 months to remove them2; CISA, the FBI and the NCSC urge everyone to follow it1.
- Patches and reboots didn't clear Cisco firewalls. In April 2026, CISA had federal civilian agencies hard-reset them9; Cisco advises reimaging10.
- Ivanti's checker can miss in-memory attacks. In October 2026, CISA warned that it may not detect exploitation of CVE-2026-104935 in Connect Secure, and that confirming it takes network and memory forensics. Connect Secure 22.8 reaches end of life on 31 January 202711.
What you can skip
- Country blocking as your main defense. Attackers route through covert networks of compromised routers that can be rapidly reshaped18. Treat geo-blocks and static blocklists as noise filters, not protection.
- A clean perimeter pen test as proof of safety. Google counted 14 zero-days in edge devices in 2025 and thinks that undercounts7. No test finds next month's flaw.
- Ripping out the VPN in a hurry. The NCSC advises keeping it until other controls give the same protection19. A ZTNA connector or cloud gateway is an edge device too, so inventory, harden and log it the same way.
Mistakes that cost the most
- Marking a device patched on a still-vulnerable version. CISA found agencies doing this with Cisco firewalls9. Check the fixed releases.
- Restoring the attacker's configuration. A post-compromise backup can bring back their admin accounts, so Fortinet and Cisco say to treat all configuration as compromised1210. Restore a checked backup or rebuild.
- Overlooking small and remote devices. Verizon found 45,000 to 50,000 end-of-life cellular modems with an exposed management interface in 2025, and saw such devices reused as attack proxies5.
What this looks like at your size
Small team, no security staff
- Ask your IT provider to list every internet-facing device with its end-of-support date, and replace any past it.
- Get free exposure alerts from Shadowserver20. In Germany, CERT-Bund alerts the operator of your IP range22; UK-based organizations can join NCSC Early Warning21.
- Check that the firewall's admin page is reachable only from inside and that VPN users need MFA.
Growing, with IT or security staff
- Forward firewall and VPN logs to your SIEM or log service, keep 12 months and alert on admin logins and configuration changes.
- Write a one-page compromise check per vendor: integrity tool, evidence to collect, reset steps and credentials to rotate.
- Add end-of-support dates to the asset inventory and budget replacements a year ahead.