Replacing the VPN with Zero Trust access
The VPN was designed to extend the office network to people's homes. It does badly a job it was never asked to do: giving dozens of people, devices and contractors measured access to a handful of applications. Here is how we replace it, without touching the rest of the network.
The problem is not the tunnel, it is what it opens
A VPN authenticates once, then places the user on the network. From there, what is reachable depends on internal filtering — which, in practice, means a great deal. A compromised personal device, a contractor account whose password has leaked, and the attacker is in the same position as an employee sitting in the office.
On top of that come everyday problems: the concentrator becomes a mandatory bottleneck that saturates at peak times, SaaS applications travel needlessly through headquarters, and nobody can say precisely who accessed what.
The Zero Trust model reverses the rule: nothing is reachable by default, access is granted resource by resource, on every connection, checking identity and device posture. Our role is to run that shift gradually, keeping the VPN as a fallback until the new path has proved itself.
- One compromised remote access opens the whole network
- Personal and contractor devices unmanaged on the internal network
- Concentrator saturated at peak times
- No fine-grained record of who reached which application
- Contractor accounts still live long after the assignment ends
An open tunnel, or a decision per resource
How we proceed
A shift in waves, application by application, with the VPN kept as a fallback throughout.
Inventory of usage
Who connects remotely, to which applications, from which devices — managed or not. That list, not the number of users, is what sizes the project.
Access policies
Defining the rules per resource: which groups, which identity conditions, which requirements on device posture (encryption, patching, active protection), and what stays outside the perimeter.
Deploying the access broker
Putting the ZTNA component in place and connecting it to the directory and to strong authentication, without exposing the applications directly on the Internet.
Migration in waves
Moving over application by application, starting with a pilot group. The VPN stays available as a fallback; it is withdrawn only once the last wave has passed and proved itself.
Control and operations
Validating the policies through a penetration test, connecting the logs to monitoring, and reviewing access periodically — contractor access in particular.
Several disciplines, a single point of contact
Designing the policies, deploying the component and checking that it holds: the design and its offensive validation are carried out by the same team.
- Inventory of remote access and the applications concerned
- Documented access policies, resource by resource
- Zero Trust access broker deployed and connected to the directory
- Plan for the migration waves and for withdrawing the VPN
- Penetration test report on the new arrangement
Frequently asked questions
Does your VPN open more than it needs to?
An inventory of remote access is enough to measure the gap between what is open and what should be.