A flat network: what most offices look like

In most small and mid-sized offices the network was built gradually. A router from the internet provider, one or two switches, a wireless access point, and then another printer, security cameras, a phone system (PBX) and a file server. Everything is connected to the same network, and every device can communicate with every other device. This situation is called a flat network.

A flat network works, which is why it is easy to ignore. The problem shows on the day something goes wrong: one computer infected with malware sees the server, the accounting team’s computers and the backup, exactly as an authorized employee sees them. Nothing stands in the way.

This situation does not come from any special negligence. It is simply the natural result of a network that grew without planning, with each new device plugged into the nearest free spot.

What segmentation is

Network segmentation is dividing the network into separate zones, with traffic between zones passing through a control point, usually the firewall. Each zone holds devices with a similar role and a similar level of trust, and between the zones it is defined explicitly what may pass.

In practice the division is usually done with VLANs: separate logical networks that run over the same switches and the same cables. This makes it possible to separate without replacing the whole physical infrastructure, provided the switches and access points support it and are configured correctly. On the wireless network, each network name (SSID) can be assigned to a different VLAN, so guests and employees connect to the same access points and reach completely different networks.

The goal is not to close everything. The goal is for every flow on the network to be one that someone decided to allow, not one that happens only because nobody blocked it.

Why it matters

The main reason is lateral movement. An attacker who got into the network, through an email with a malicious file (phishing), a leaked password or a device that was not updated, almost never settles for the first computer. They scan the network, look for servers, admin accounts and backups, and move forward. In a flat network the way is open. In a segmented network they run into a firewall at every crossing, and every such attempt is also a chance to detect them.

There are everyday reasons too. Devices such as cameras, printers, meeting room screens and air conditioning controllers (IoT) receive security updates rarely, if at all. Guests and suppliers connect to the Wi-Fi with computers the business has no control over. Separation ensures that a weak or foreign device does not sit on the same network as the business’s sensitive data.

And finally, regulation. The National Cyber Directorate guidance and the ISO 27001 standard explicitly address network separation, and the Privacy Protection Regulations (Data Security), 2017 under the Privacy Protection Law require limiting access to databases of personal data (access control). An organization that holds data about customers or employees needs to be able to show that access to this data is actually limited. A written procedure alone is not enough. When an incident occurs, the question of what was separated from what comes up very quickly, with the Privacy Protection Authority, the insurance company and customers.

Which zones to separate in a typical office

There is no single division that fits everyone, but in most offices the same zones come up:

  • Workstations (endpoints): the employees’ computers. This is the zone with the highest risk of infection, because this is where email is opened and browsing happens.
  • Servers: the file server, Active Directory and specialized systems. They are reached only over the services that are needed, not over every protocol.
  • Backup: a zone that almost nobody needs to access. The more isolated it is, the smaller the chance that ransomware will reach it too.
  • Management (management network): the management interfaces of the switches, the firewall, the physical servers and the virtualization system. They should not be reachable from every computer on the network.
  • Smart devices (IoT): cameras, building controllers and screens. They usually need very limited access, sometimes only to one server.
  • Printers and telephony (VoIP): one or two separate zones, depending on the size of the office.
  • Guests (guest network): internet access only, with no path to the internal network.

A small office can start with three or four zones and add more later. It is more important for the division to be documented and clear than for it to be detailed.

The rules between zones are what count

The division itself is only the infrastructure. What determines the level of protection is the rules set in the firewall between the zones. The common approach is blocking by default (default deny): nothing passes between zones except what was explicitly defined. Workstations reach the file server and the systems they need, management interfaces are reachable only from designated management stations, and the guest network goes out to the internet only.

It is also worth logging and monitoring what gets blocked (logging). Repeated attempts from a regular computer to reach a management interface or the backup server are a sign that something is wrong, and should be checked right away.

How to do it without stopping the office

Moving from a flat network to a segmented one touches almost every device in the office, so it is done in stages:

  1. Mapping (asset inventory): a list of all devices, what each one does and what it needs to communicate with. This stage usually reveals devices nobody remembered.
  2. Planning: the division into zones, address ranges (subnets), and a table of what may pass from zone to zone.
  3. Checking the infrastructure: whether the switches, access points and firewall support the division and the load it adds.
  4. Gradual migration: start with the simple zones, such as guests and smart devices, and leave the workstations and servers for last, outside working hours.
  5. Monitoring and tightening: in the first weeks, check what is being blocked and what is missing, and narrow down rules that are too broad.

Throughout the process it is important to document. An up-to-date network diagram and an explained rule table are what make it possible for whoever comes along a year later to understand why each rule exists.

Who plans and who maintains

Segmentation is not a one-time project. Every new device, new system or supplier that gets remote access changes the picture. If the rules are not maintained, after a year or two temporary exceptions that became permanent pile up, and the network is flat again in practice.

That is why whoever manages IT for the business also needs to understand information security. An IT company that sees the network as a matter of connectivity alone will build a network that works. A company that knows the attack methods (attack vectors), the regulatory requirements and the National Cyber Directorate guidance will build a network that holds up even when something goes wrong. The difference shows exactly on the day one computer is infected: whether the incident stops there, or the whole office is shut down and customer data is exposed.

Summary: where to start

  • Check whether your network is flat: whether a regular computer can see the servers, the backup and the management interfaces.
  • Separate the guest network and the smart devices first. That is the simplest step.
  • Isolate the backup and the management interfaces.
  • Set blocking by default (default deny) between zones, and document every exception.
  • Decide who is responsible for updating the rules when something changes.