Skip to content
Tutorials Center
My learning

Topics

Lesson 26 of 37 ยท Wireless and Zero Trust

Zero Trust Principles

Learn what zero trust means: never trust, always verify, least privilege and assume breach, plus the policy engine, identity, device posture and micro-segmentation behind it.

The classic network model is a castle with a moat. Outside is dangerous, inside is trusted. Once a laptop sits on the office LAN or a user connects through the VPN, it can usually reach a great deal. That model broke down for three reasons: attackers who get one foothold (a phished laptop) move freely inside, staff work from anywhere, and applications now live in clouds the perimeter does not enclose.

Zero trust is the response. Its slogan is never trust, always verify: no user, device or network location is trusted by default. Every request to every resource is authenticated, authorized and evaluated against current context, whether it comes from the office or a coffee shop. Zero trust is not a product you buy; it is an architecture and a set of habits, described in standards such as NIST SP 800-207.

The core ideas

Never trust, always verify

Verify identity Strong login, MFA
Verify device Patched, managed, healthy
Verify context Time, place, risk

Network location alone proves nothing.

1 / 3
  1. Verify explicitly. Decide each access using every available signal: who the user is, which device, how healthy it is, where the request comes from, what is being accessed, and how risky the behavior looks.
  2. Use least privilege. Grant the minimum access, for the minimum time (just-in-time elevation), and review it. Role- and attribute-based access control are the usual ways to express it.
  3. Assume breach. Segment so that a compromised account or host cannot roam, encrypt internal traffic, and monitor continuously as covered in log analysis.

Architecture: policy engine and enforcement point

NIST describes a split between deciding and enforcing:

flowchart LR
    U["User + device"] --> PEP["Policy Enforcement Point<br/>(proxy / gateway)"]
    PEP -- "ask" --> PDP["Policy Decision Point<br/>(engine + administrator)"]
    PDP --- ID["Identity provider"]
    PDP --- DV["Device posture"]
    PDP --- TI["Threat signals / logs"]
    PEP -- "allowed" --> R["Resource (app, API, server)"]

Every request goes through an enforcement point that asks the decision point for a verdict.

  • The policy decision point (PDP) evaluates rules using inputs from the identity provider (who), device posture (the state of the machine: patched, disk encrypted, endpoint agent running), and threat intelligence and logs (is this account behaving oddly?).
  • The policy enforcement point (PEP) is a gateway, reverse proxy or agent that sits in front of the resource and allows, denies or limits sessions based on the decision.
  • Resources are not reachable directly. In a real deployment the application is hidden behind the PEP and has no open inbound ports on the internet.

The next lesson builds a small policy engine so you can see these decisions in code.

Zero trust in practice: building blocks

  • Strong identity. Single sign-on with phishing-resistant MFA (passkeys or hardware keys). Identity becomes the new perimeter.
  • Device trust. Only managed, patched, encrypted devices reach sensitive systems; the posture is re-checked continuously.
  • ZTNA (Zero Trust Network Access). Replaces the full-network VPN with per-application access: a user can reach the HR app but is not placed "on the network" at all. This contrasts with the classic VPN model that grants broad reach once connected.
  • Micro-segmentation. Workloads are isolated into small zones using host firewalls, cloud security groups or software-defined networking, with explicit allow rules between them. East-west traffic (server to server) is controlled just like north-south.
  • Encryption everywhere. TLS or mutual TLS between services, so the internal network is not assumed private (secure protocols).
  • Continuous monitoring and automation. Trust is evaluated per request, not granted once; a risky signal can step up authentication or cut a session.

A migration path, not a switch

Zero trust is adopted gradually:

  1. Inventory users, devices, applications and data flows. You cannot protect what you cannot list.
  2. Strengthen identity: MFA everywhere, remove shared and stale accounts.
  3. Protect the most valuable applications first with per-app access.
  4. Segment the flat internal network step by step, starting in monitor mode to see which flows really exist.
  5. Measure and refine, using logs of denied requests to tune policy.

Common misconceptions

  • "Zero trust means trust nobody and block everything." It means verify everything, then allow what policy permits, with a smooth user experience.
  • "We bought a zero-trust product, so we are done." Zero trust spans identity, devices, network, applications and data.
  • "VPNs are dead." They are often replaced or reduced, but remain in transition, and legacy systems may still need them.
  • "It only matters for remote work." Insider threats and lateral movement are internal problems too.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What does 'never trust, always verify' mean?

    Each request is evaluated using identity, device and context, regardless of where it originates.
  2. 2. Which component enforces the decision in a zero trust architecture?

    The PEP (a gateway or proxy) allows or blocks sessions based on what the policy decision point decides.
  3. 3. How does ZTNA differ from a traditional VPN?

    ZTNA brokers per-application access, so a connected user cannot roam the whole internal network.
  4. 4. What does the 'assume breach' principle drive you to do?

    If you expect an attacker to get in somewhere, you design to limit and detect what they can do.

Summary

  • Zero trust replaces perimeter trust with per-request verification of identity, device and context.
  • Pillars: verify explicitly, least privilege, assume breach.
  • A policy decision point evaluates signals; a policy enforcement point allows or denies.
  • Building blocks: strong identity, device posture, ZTNA, micro-segmentation, encryption, continuous monitoring.
  • Adopt gradually: inventory, identity, protect key apps, segment, measure.

Next: Building a Zero Trust Policy Engine