(Hi there readers – If you know anyone who works for an SSL VPN vendor – in development, product management and so on, please send them a link to this post, hoping this idea will gain attention, acceptance, and eventually will come to life, hence enhance security for all of us)

The Hebrew version of this post can be found here.

TL;DR / Summary

This is an idea I came up with, to enhance the security of SSL VPN, by reducing the SSL VPN server attack surface, by adding Dynamic DNS (DDNS) client to the SSL VPN client endpoint, hence configuring the server side to allow only networking-pre-approved sources to even communicate with the SSL VPN server component of the gateway in the first place, based on FQDN values of the DDNS clients, hence blocking, up front, most of the internet from even communicating with the SSL VPN server, hence greatly reducing its attack surface.

Thinking about it further – the above principle is good for any server element who needs to accept incoming traffic non-fixed source IP(s) – be it a web server, an SSH server and so on

I tried so find solutions that use the above principle, but I didn’t find any.

From what I have found – all uses of a DDNS client are for the benefit of systems that operate in as a server, meaning they listen for client-initiated incoming traffic, in order to provide service to clients, so that clients can find them and reach them, even if the IP address of these systems changes.

There are security gateway products who support using FQDN as an object in their Firewall rules, which means the above idea can be technically realized already now, but I think it will be hard to manage it all this way, and it will not scale. I think this needs a more systematic solution, to also handle the client side and DDNS server side, so I think vendors should incorporate it as part of their relevant products and services, to be able to be managed properly by customers.

Background

I noticed that in the recent time there are many incidents in which SSL VPN components of security gateways are breached successfully, due to various reasons, mostly due to software vulnerabilities of the gateway’s software.

Vulnerabilities are inevitable, it will always happen, as we cannot guarantee bulletproof, 100 percent secure, software.

Also, SSL VPN, as a concept, is mostly used to grant secure access from the Internet in to local, private, internal networks, so it has to be open to the whole (or most) of the internet, as the organization cannot know in advance from which source IP addresses the clients will initiate communicate to the gateway, and as also some of the client endpoints change their source  IP from time to time, for various reasons (Their ISP changed their allocated IP, they travel in train and change cellular networks when they cross to another country, and so on).

So, I thought, we have here an enormous attack surface at the gateway side, constantly open to all of the Internet, so no wonder it is successfully breached so much.
I though to myself: There should be a better, more secure, way to do it.

Here are several news articles about mostly recent SSL VPN vulnerabilities and breaches (sorted from newest to older):

2025, August
Fortinet SSL VPNs Hit by Global Brute-Force Wave Before Attackers Shift to FortiManager
https://thehackernews.com/2025/08/fortinet-ssl-vpns-hit-by-global-brute.html

2025, August
SonicWall Investigating Potential SSL VPN Zero-Day After 20+ Targeted Attacks Reported
https://thehackernews.com/2025/08/sonicwall-investigating-potential-ssl.html

2025, April
Ivanti has recently patched a critical severity vulnerability found in its Connect Secure (ICS) VPN appliances which was allegedly being abused in the wild by Chinese state-sponsored actors
https://www.techradar.com/pro/security/ivanti-patches-serious-connect-secure-flaw

2025, March
VPN Vulnerabilities Become a Primary Weapon for Threat Actors Targeting Organizations
https://gbhackers.com/vpn-vulnerabilities-become-a-primary-weapon/

2025, January
Ivanti Connect Secure VPN Targeted in New Zero-Day Exploitation
https://cloud.google.com/blog/topics/threat-intelligence/ivanti-connect-secure-vpn-zero-day

2024, February
Exploitation Observed: Ivanti Connect Secure — CVE-2023-46805 and CVE-2024-21887
https://www.akamai.com/blog/security-research/ivanti-january-rce-cve-zero-day-exploitation-observed

2023, July
Researchers Develop Exploit Code for Critical Fortinet VPN Bug
Some 340,000 FortiGate SSL VPN appliances remain exposed to the threat more than three weeks after Fortinet released firmware updates to address the issue
https://www.darkreading.com/vulnerabilities-threats/researchers-develop-exploit-code-for-critical-fortinet-bug

2022, December
Fortinet Warns of Active Exploitation of New SSL-VPN Pre-auth RCE Vulnerability
https://thehackernews.com/2022/12/fortinet-warns-of-active-exploitation.html

The suggested solution (High level)

The core concept for using DDNS is that if we wish to grant secure access to clients that we cannot know in advance what is their source IP and even if we know it at some point – it can change at any time, because it is not a fixed VPN connection, with fixed client IP, so we use the DDNS system to have a trusted fixation layer (made of the FQDN’s domain and DDNS server), which we trust to authenticate the client device – hence we are willing to allow the first networking level access to an unknown-in-advance client IPs, but ones that have a preliminary required software (the DDNS client) and was authenticated (even if just a device was verified, not a human) at a system we trust (The DDNS server system).

The idea is to limit who has access to the SSL VPN server side, based on the following high-level concept:

• SSL VPN Client side
  • The clients who wish to access and use the SSL VPN server – will have, as part of their SSL VPN client software (and possibly as part of their EDR client software), a Dynamic DNS (DDNS) client component
  • The DDNS client will communicate securely with its DDNS server (I guess over HTTPS), authenticate to it, and send to the DDNS server its client identity and public IP address (which the DDNS server can also fetch by itself, from the network session). This should be done every time the public IP of the client changes, to verify the DDNS server and the gateway always get the current IP of the client
    • The DDNS client software must of course be secured – The client’s DDNS server credentials should be encrypted locally; communication to the DDNS server needs to be encrypted and authenticated, DNS queries better be encrypted using DoH (DNS over HTTPS); protected from tampering and so on

• SSL VPN server side (the organizational security gateway)
  • The gateway will, constantly and ad-hoc when needed, run reverse (PTR) DNS queries against the same DDNS server/system as the SSL VPN client use, to fetch the current IP addresses of relevant SSL VPN clients
  • The gateway will have a firewall rule, related to the SSL VPN activity, who allows access to it from the internet, based not on numeric source IPs, but rather on FQDN names that the DDNS server controls and manages, the ones attached to its clients (desired with wildcard support, for better configuration flexibility)
    • For example, the Firewall rule’s allowed source object will be *.vpn-allowed.firm.com, and any IP that is resolved to matching FQDN, will be allowed (e.g., johndoe.vpn.firm.com or serioussam.vpn.firm.com)

• Session
  • The client will initiate an https-based SSL VPN session to the gateway
  • The gateway will receive the first packet of the TCP handshake session (with the SYN flag) and will check the value of the incoming source IP, based on the above SSL VPN access firewall rule
  • This source IP will be evaluated (best using a DNS PTR resolving (hence IP to FQDN) against the DDNS server or, less desired, as its data may not be up-to-date, locally against a pre-fetched cache)
    • Timeouts and invalidation of non-active DDNS client’s is highly desired, to prevent granting access to unauthorized clients who accepted the original IP if the real client is not using that IP anymore, for any reason
  • If there is a match – the client will be allowed to continue with the session – to complete the TCP handshake, set the TLS session, authenticate and then establish the SSL VPN connection
  • If there is no match – the firewall rule evaluation process will continue to the rest of the firewall’s rules, ideally eventually reaching the last rule, the cleanup / deny-all rule, which will drop the communication, hence blocking any further communication from the client side to the SSL VPN server, which actually means blocking any source IP that does not currently has a currently active allowed DDNS client ID, which is probably most of the Internet, which means most of the opportunistic attackers

Added possible points:

• The DDNS server system
  • I think it is desired not to use a DDNS server system that is part of the gateway, the SSL VPN server device – as it is exposing the device to the whole Internet using an extra web/DNS server interface, which is against what we try to achieve here in the first place
  • Also, I think it is less desired to use 3^rd party DDNS system/service, unless it is a very secure one
  • Ideally the DDNS system will be a cloud service by the same vendor of the security gateway’s vendor, to hopefully achieve more compatibility, interoperability, and security
• Of course, this concept adds to the current way of things – more complexity, management, and the need for hardening, but I think the added security worth it

Zero trust features that can be added as well:

• Applying IP reputation – to either block access up front, or instead adding more security checks or limitations to the session
• Dropping the session if the public IP of the client changed during an active session, forcing the client to start a new session from the beginning
• Blocking clients if their public IP is changing too fast/often or their public IP is not reported within a timely manner (if there is a technical demand from the DDNS logic for recuring scheduled updates from the DDNS client to its server)
• Checking local settings like if the client is set as router (IP Forwarding), the use of a proxy, the use of another VPN in parallel, and so on (although this slides more towards an EDR functionality)

That’s all, folks. I hope you liked it, and if you did – please spread the word, and the link to this post, to relevant people who can bring this concept to life.