What regular segmentation does not stop
Dividing the network into zones, workstations, servers, management and guests, is an important step. But it mainly handles traffic between zones. Inside a single zone, all devices still see each other. Dozens of workstations on the same VLAN can communicate with one another, and servers in the same zone reach each other over any protocol.
That is exactly the space in which an attacker advances after gaining a foothold. Modern ransomware does not encrypt the first computer and stop. It looks for passwords, connects to neighboring computers, reaches a server and from there everything it can. The difference between an incident that hit one workstation and an incident that shut down the organization is often decided in the first hours, by how far the attacker managed to get.
What micro-segmentation is
Micro-segmentation takes separation down to the level of the server, the virtual machine (VM) and sometimes the application. Instead of a rule that says “the workstation zone reaches the server zone”, the rule says “the accounting team’s computers reach the accounting server, on the application’s port only”. Everything else is blocked.
This is a practical application of the Zero Trust principle: a device is not trusted just because it is on the internal network. Every connection needs a justification, and anything that was not defined is blocked. The result is that an incident on one computer stays far more limited, because that computer simply does not reach most of the systems in the organization.
It is important to say that micro-segmentation does not replace regular separation. It is another layer on top of it (defense in depth), and it works best when the basics are already in order: clear zones, a firewall between the zones and documentation of the network.
What lateral movement looks like in practice
To understand why this matters, it helps to know the recurring pattern. An attacker gains access to one workstation, usually through a phishing email or a leaked password. From there they collect sign-in details saved on the computer (credential dumping), scan the network and look for devices that accept management connections such as RDP, SMB or remote PowerShell. Every computer that accepts such connections from any other computer is another step on the way to the servers and the admin accounts.
In most organizations there is no business reason for one workstation to connect by remote desktop to another workstation, or to reach file shares on other employees’ computers. Blocking this traffic alone closes off a large share of the routes attackers use, and is hardly noticed in daily work.
It is important to remember that an attacker does not need special hacking tools. Much lateral movement is done with tools and protocols that system administrators use every day (living off the land), so it does not always look suspicious. Limiting who may use them, and from which computer, is one of the effective ways to stop it.
Ways to implement it
There are several approaches, and many organizations combine them:
- A local firewall with central management (host-based firewall): the Windows firewall or the operating system firewall on Linux servers, with rules defined and managed centrally, for example through Group Policy (GPO). This is a starting point available in almost every organization.
- A firewall at the virtualization level (distributed firewall): in virtual environments it is possible to filter traffic between virtual machines even when they run on the same physical server and on the same network.
- Dedicated platforms: systems that install a component on every server and workstation, map the actual communication and make it possible to define policy by role rather than by IP address.
- Capabilities of the EDR system: some EDR systems include management of the local firewall and immediate isolation of a suspicious computer from the network (host isolation).
The choice depends on the size of the environment, the existing systems and the ability to maintain the policy over time. A sophisticated solution nobody updates is worse than a simple solution that is maintained.
Where to start
Micro-segmentation across the whole organization at once is a recipe for failures. The right way is to start with the steps that bring the most benefit and with the most important systems:
- Blocking communication between workstations: most businesses do not need employees’ computers to talk to each other directly.
- Limiting management protocols: RDP, administrative shares (admin shares) and remote PowerShell are allowed only from designated management stations (privileged access workstation).
- Protecting the most important assets (crown jewels): Active Directory, the backup server, financial systems and databases of personal data. For each of them it is defined explicitly who reaches it and on which ports.
- Gradual expansion: application by application, with documentation of every approved flow.
Even after it is turned on, the policy needs maintenance. A new system, a replaced server or a supplier that was given access require the rules to be updated. When the update is not done, pressure builds to add broad exceptions so work can continue, and after a while the protection wears down. So it is worth setting an orderly process (change management): every request for a new rule is documented, gets a justification and a date for review.
The identity side
Network separation alone is not enough if an attacker gets an admin account that is allowed to reach everywhere. That is why micro-segmentation goes hand in hand with identity and access management (IAM):
- LAPS: a unique password for the local administrator account on every computer, so that one stolen password does not open every computer in the organization.
- Separate admin accounts: system administrators work day to day with a regular account, and use an admin account only when needed and only from a dedicated station.
- MFA on all remote access and on privileged accounts.
- PAM in larger organizations: central management of protected access, with a record of who connected where and when.
The combination of restrictions in the network and restrictions on identities is what makes lateral movement truly hard. Either one alone leaves an opening.
Why this takes a provider who understands security
Micro-segmentation sits exactly where IT operations and information security meet. Whoever defines the rules needs to know the organization’s systems in depth, which services each server provides and what dependencies exist between them. They also need to understand how attackers move through a network, and what is required by the National Cyber Directorate guidance and the Privacy Protection Regulations (Data Security), 2017 under the Privacy Protection Law.
When the business’s IT company understands both sides, the separation is built correctly from the start and maintained as part of the ongoing work. When it does not, the organization is left with a network that looks organized on paper, and discovers the gap only when an incident spreads beyond the first computer, stops work and reaches customer data.
Summary
- Segmentation between zones is the basis, and micro-segmentation closes off traffic inside the zones (east-west traffic).
- The first step that pays off: block direct communication between workstations and limit management protocols.
- Protect the most important assets first: Active Directory, the backup and the databases.
- Monitor before blocking (monitor mode), and document every approved flow.
- Combine with LAPS, separate admin accounts and MFA.