Network segmentation

We split a flat corporate network into zones so that one infected device cannot reach the entire network.

When you need it

When one workstation can reach everything

In most organizations the internal network is flat: from a single workstation you can reach the servers, the backup system, the printers, and every other workstation. This causes no trouble for years, right up to the first incident.

  1. The attack starts on the inside

    A typical incident begins with phishing or a stolen password, so the attacker gets around the perimeter defenses. On a flat network, moving from the first compromised machine to the real target costs them very little.

  2. The network decides how much damage is done

    What matters is how many systems the attacker can reach from the first machine, how quickly they gain higher privileges, and whether they can get to the backups. All three depend on how the network is built, and endpoint protection does not help with that.

  3. VLANs, but no boundary

    A VLAN is a logical separation on a single physical infrastructure. If traffic is routed between VLANs without filtering, the segments can reach each other freely, so the zone boundary needs a decision point.

Zone model

What to separate first

The zone model records which system goes into which zone. Each of the four zones below can be separated on its own. Outbound traffic is part of the model too, because restricting it is what stops an attacker from setting up a command-and-control channel and from exfiltrating data.

Segmenting the internal network: where to start
  1. Backup infrastructure

    If the backup server can be reached from the same segment as the workstations, ransomware deletes the backups first and encrypts afterwards. So backups go into a separate segment that accepts traffic only from the systems being backed up.

  2. Domain controllers and identity

    Domain controllers get their own segment with strictly defined inbound traffic. The certificate authority (ADCS), which is often left out, belongs here too. Administrative access is possible only through a jump server that is isolated as well.

  3. Workstations isolated from each other

    With client isolation, machines in the user segment cannot reach each other, only the servers and the internet egress. Most managed switches support it, and it is exactly what stops attacks that spread from machine to machine.

  4. Unmanaged devices

    Printers, IP cameras, access control systems, and building automation devices often run with default passwords and firmware that has never been updated. They go into their own segment with restricted outbound traffic, and their management interfaces can be reached only from the IT segment.

The process

How we split the network into zones

Many organizations are wary of segmentation because the first faulty rule breaks something, and from then on nobody is allowed to touch the network. That is why we schedule the rollout and plan it so that it does not disrupt operations.

  1. Mapping

    We document the current network and look at what talks to what. NetFlow or the firewall log is enough to collect the traffic. Traffic first, rules second: a rule written from the documentation alone will always have gaps.

  2. Zone model and rules

    We define the zones and design the traffic allowed between them. We follow the principle of least privilege: anything not explicitly allowed is denied.

  3. Scheduled rollout

    We roll out the zones and the rules between them on a schedule, so that the rollout does not disrupt operations.

  4. Final test and handover

    We check that the separated systems can be reached from other zones only in the ways the rules allow. We document the final zone model and rule set and hand them over.

Deliverables

What you get at the end

items handed over

  • Documentation of the current network
  • Zone model and rules between zones
  • Scheduled rollout that does not disrupt operations
  • Final test and documentation

Frequently asked questions

The questions we hear most

How can I tell whether our network is flat?

With a quick self-test: sit down at an ordinary workstation and try to open the admin interface of the backup server. If the login screen appears, and it does surprisingly often, segmentation is missing.

Should we start with microsegmentation?

Usually not. Microsegmentation splits the network into zones at the application or workload level, with strictly defined traffic. It is the strongest barrier against attacks that spread from machine to machine, but it takes a lot of design work and delivers only a fraction of the benefit you get from separating a few well-chosen zones. It is better to start with the basic zones and go further later. In our Zero Trust guide, application-level segmentation is the third round.

Is this the right service for an industrial network too?

We have a separate service for OT environments. There we split the flat OT network into zones so that a compromised device cannot be used to spread across the whole plant. The zones and the conduits between them follow the IEC 62443 model. Details are on the OT network segmentation page.

What needs to happen after the rollout?

The rule set needs regular review. Most firewalls show how many times each rule has been hit, and rules with zero hits can be deleted. If nobody records why a line is in the rule set, no one dares to touch it later, and from then on it only grows. We also take on periodic rule set reviews as network maintenance, under a custom agreement and within a scope agreed with the client.

Contact

Ready to secure your critical systems?

Our experts will assess where your security stands and build a roadmap that fits your organization.

ARLITECH's office in Gödöllő

Visit us

You are welcome to visit our Gödöllő office in person.

Address
Centrum Office Building, 2100 Gödöllő, Dózsa György út 13, 3rd floor, office 309
Office hours
Mon–Fri, 9 a.m. to 5 p.m.
Get directions (opens in a new tab)