The VikiD Truth

Knowledge base / Computing / Security

Read This! | Picolisp | Picolisp Machine | Pil Sources | Linux | BASH | C-Programmming | Operating Systems | Computer Security | ICSsecurity | Firewalls | Networking | Security Programs | Exploitation | Social Engineering | Metasploit | wifi-penetration | Forensics | Secure Network Architecture

Secure Network Architecture

Principles

  1. Isolate ICS (Control) Network & Corporate (Production) Network
  2. Use minimal connections between Control & Production Networks
  3. use Firewalls
  4. Filter Packets - ICMP codes, especially Destination Unreachable Network & Host flags
  5. Block all Communication by Default
  6. Enforce Secure Authentication
  7. Enforce Destination Authorisation
  8. Record Information
  9. Permit the ICS to Implement Operational Policies

Netwok Architecture

Logically Seperated Control Networks

Defense in Depth / Onion Security

  1. Application & Data
  2. Data End Point
  3. Patch Management - Qualys Guard
  4. Intrusion Prevention
  5. Virus Protection
  6. Host-Based Firewall
  7. Server Hardening
  8. Internal Network
  9. Perimeter Security
  10. Physical Security
  11. Policies & Procedures

Use Secure Protocols

The plain idea

Secure network architecture is the art of designing a network so that a break-in anywhere does the least possible damage. The single biggest idea, running through every principle above, is segmentation: split the network into zones, keep the connections between them few and tightly controlled, and deny everything that is not explicitly allowed. My examples focus on the classic and vital case of keeping an industrial control network separate from the ordinary corporate network.

Why it matters

A flat network, where everything can talk to everything, is a house with no internal doors: one compromised laptop and the attacker can reach the servers, the backups and the factory floor. Segmentation turns that into a house of locked rooms. It is why the 2016 Ukraine power-grid attack and countless ransomware outbreaks spread so far, and why zoning is the first thing any serious review looks for.

An everyday analogy: bulkheads in a ship

A ship’s hull is divided into watertight compartments. Hole the bow and one compartment floods, but the bulkheads stop the water reaching the rest, and the ship stays afloat. Network segments are bulkheads: a breach in one zone is contained instead of sinking everything.

A worked example: reading the diagrams

My three sketches are three strengths of the same idea:

  1. Dual-homed host. One machine with two network cards bridges two zones. Cheap, but that single host is now the only thing between them, a weak bulkhead.
  2. Firewall between zones. A dedicated firewall enforces the rules, far stronger because its only job is to police what crosses.
  3. Firewall and router. Adds a routed boundary and a second checkpoint. The strongest of the three, and the basis of the modern DMZ (demilitarised zone), where internet-facing services sit in a middle zone that can reach neither the internet nor the inside freely.

The “filter ICMP, especially Destination Unreachable” principle is a nice detail: those messages leak network structure to an attacker mapping you (see enumeration), so a careful design quietens them.

A 2026 update

The frontier today is zero-trust architecture: stop trusting anything just because it is “inside” the network, and instead verify every request by identity and device, every time. Combined with micro-segmentation (segments as small as a single workload) it is segmentation taken to its logical end. For control systems, the Purdue model of layered zones and a DMZ between IT and OT remain the reference design.

Common misconceptions

How it connects

The truth revealed: you cannot stop every break-in, so the real skill is designing the network so that one break-in is never the whole story.

Last edited · 27 revisions

Tags: #Computing