Network Segmentation and Defense in Depth

Network Segmentation and Defense in Depth

6 min read4 views

Super AdminSuper Admin

A flat network is convenient right up until something on it is compromised. Segmentation is the practice of dividing a network into zones so that a breach in one place does not become a breach everywhere, and it is one of the most practical expressions of defense in depth.

This article explains the main segmentation techniques, how they fit into a layered security model and how the same ideas lead naturally towards zero trust.

Defense in depth in one paragraph

Defense in depth assumes that any single control can fail. Instead of relying on one strong perimeter, you place several independent layers between an attacker and what you value, so that each layer either stops the attack or slows it and makes it visible. The layers should differ in kind, technical, administrative and physical, because controls that share a weakness fail together.

Layered controls and what each layer contributes
LayerExample controlsWhat it helps stop
Policies and peopleAcceptable use policy, security awareness trainingPhishing success, unsafe practices
PhysicalBadge access, locked racks, CCTVDirect access to hardware and ports
PerimeterEdge firewall, DDoS protection, email filteringUnsolicited inbound traffic, bulk attacks
Internal networkSegmentation, internal firewalls, NAC, IDS/IPSLateral movement, rogue devices
HostHardening, EDR, host firewall, patchingMalware execution, exploitation of local flaws
ApplicationSecure coding, WAF, strong authenticationInjection, account takeover
DataEncryption, classification, DLP, backupsExfiltration, loss, ransomware impact

Segmentation lives mainly in the internal network layer, but it multiplies the value of the others: monitoring is far more effective when traffic between zones is expected to be rare and well defined.

Two engineers walking along an aisle in a data centre
Photo: Unsplash

Segmentation techniques

The techniques below range from coarse to fine-grained. Most real environments use several at once.

VLANs

A virtual LAN splits one physical switch infrastructure into separate layer 2 broadcast domains. Devices in different VLANs cannot talk directly; traffic between them must pass through a router or layer 3 switch, which is where you can apply policy.

VLANs are an organizing tool, not a security control on their own. Know the classic VLAN hopping attacks: switch spoofing, where an attacker negotiates a trunk link, and double tagging, which abuses the native VLAN. Mitigations include disabling automatic trunk negotiation on access ports, setting the native VLAN to an unused ID and not placing user devices in VLAN 1.

ACLs and firewalls

Access control lists on routers filter traffic by address, protocol and port, and are stateless: each packet is judged alone. Stateful firewalls track connections, so return traffic for an allowed session is permitted automatically. Next-generation firewalls add application awareness and user identity.

The guiding principle is default deny: write explicit allow rules for the flows the business needs, and let everything else fall to a final deny rule that is logged. The illustrative rule set below, in plain pseudo-syntax rather than any vendor's format, shows the idea for a three-tier web application.

# Zones: INTERNET, DMZ, APP, DATA, MGMT
# Rules are evaluated top to bottom; the first match wins.

allow  INTERNET      -> DMZ:web-proxy    tcp/443
allow  DMZ:web-proxy -> APP:api          tcp/8443
allow  APP:api       -> DATA:database    tcp/5432
allow  MGMT:bastion  -> APP, DATA        tcp/22          log
allow  APP, DATA     -> INTERNAL:dns     udp/53, tcp/53
deny   DMZ           -> DATA             any             log
deny   any           -> MGMT             any             log
deny   any           -> any              any             log   # default deny

Notice that the internet can only reach the proxy, the proxy can only reach the API, and only the API reaches the database. Administrative access arrives exclusively through a bastion host. Even if the web proxy is compromised, the attacker has no direct path to the data tier.

The DMZ

A demilitarized zone, or screened subnet, holds systems that must accept connections from untrusted networks, such as public web servers, reverse proxies and mail gateways. It sits between an external and an internal firewall, or on a separate interface of one firewall. The rule of thumb: traffic may flow from the internet into the DMZ and from the internal network into the DMZ, but connections initiated from the DMZ into the internal network should be rare, specific and closely monitored.

Microsegmentation

Traditional segmentation draws boundaries around groups of systems. Microsegmentation draws them around individual workloads, often enforced by host-based agents, hypervisor features or cloud security groups. Policy is written in terms of identity and role ("payment API instances may reach the payment database") rather than IP ranges, and it follows workloads as they move or scale.

This is where segmentation meets zero trust. Zero trust removes implicit trust based on network location: every request is authenticated, authorized and evaluated against policy, wherever it comes from. Microsegmentation provides the enforcement points that make that possible inside the data centre. NIST describes the architecture in SP 800-207.

Why segmentation matters: lateral movement

Hypothetical scenario: An attacker phishes an employee in the finance department and gains a foothold on their laptop. On a flat network, that laptop can reach the file servers, the domain controllers and the backup system directly, so the attacker scans, harvests credentials and moves sideways until they find the backups and encrypt them. On a segmented network, the finance VLAN can reach only the services finance needs; administrative protocols to servers are blocked except from a hardened jump host; and the backup network accepts connections only from the backup server. The initial compromise is the same, but the blast radius is a single workstation, and the blocked connection attempts generate alerts.

That is the real payoff. Segmentation rarely stops the first step of an attack; it stops the second, third and fourth, and it buys detection time.

Earth at night with glowing points of connection across the globe
Photo: Unsplash

Designing and maintaining segments

Good segmentation starts with understanding the business, not the switches. A practical approach:

  1. Inventory assets and classify them by sensitivity and function.
  2. Map the flows that genuinely need to happen, ideally from observed traffic as well as interviews.
  3. Group assets into zones with similar trust and requirements, such as user devices, servers, payment systems, management and guest Wi-Fi.
  4. Write default-deny policies between zones with narrow allow rules.
  5. Monitor denied traffic, review rules on a schedule and remove anything nobody can justify.

Rule sets decay. Temporary exceptions become permanent and "any" rules creep in under deadline pressure, so periodic review is part of the control, not an optional extra. Segmentation also reduces compliance scope: isolating cardholder data systems, for example, can limit how much of the environment a PCI DSS assessment must cover.

What the exams expect

Network+ focuses on mechanics: VLANs, trunking, ACLs, DMZs and firewall types. Security+ and CISSP ask why: which design limits lateral movement, where a public web server belongs, or which principle a default deny rule implements. When in doubt, choose the answer that reduces implicit trust and narrows each permitted path to the minimum.

Conclusion

Segmentation turns one large target into many small ones. VLANs organize, firewalls and ACLs enforce, the DMZ isolates what must face the internet, and microsegmentation carries the same logic down to individual workloads in a zero trust design. Combined with the other layers of defense in depth, it means a single compromise becomes an incident you contain rather than a breach you explain.

0.0 (0 ratings)
Share:

Network+

1,842 practice questions

How ready are you for Network+?

Take the entry assessment and see exactly which domains need work.

Take the assessment

24 questions · about 15 minutes · no account needed

Discussion (0)

No discussion here yet — be the first to comment.

Please sign in to join the discussion.

More posts on Network+