DOMAIN 3 · Security Architecture and Engineering
Secure design principles and zero trust
Apply least privilege, explicit authorization, separation of responsibilities, and layered controls to concrete architecture decisions.
Turn principles into decisions
Section titled “Turn principles into decisions”A design principle is useful when it changes a boundary, permission, default, or failure behavior. Begin with the resource to protect and a credible misuse scenario. Then identify who requests an action, which component decides, where that decision is enforced, and what evidence is recorded.
ISO’s public overview identifies access control and cryptography among the subjects of ISO/IEC 27002 control guidance.1 The examples below are original learning explanations, not reproductions of ISO controls.
Limit privilege and scope
Section titled “Limit privilege and scope”Least privilege gives a subject only the permissions needed for its task. Apply it to people, service identities, deployment systems, and administrative tools. Separate an application’s read-only reporting identity from its account-management identity so that compromising one function does not automatically expose the other.
Time and scope matter as much as action names. Temporary administrative access can be narrower than a permanent administrator account, but it still needs approval, enforcement, and revocation. A permission listed in documentation is ineffective if the runtime identity retains broader access.
Explain zero trust precisely
Section titled “Explain zero trust precisely”NIST SP 800-207 treats network location as insufficient grounds for trust. Its tenets include securing communications regardless of location, granting access per session, using dynamic policy, and monitoring asset security posture.2
This does not mean all requests must be denied. It means authorization should be based on appropriate evidence and limited to the requested resource. A device on an office network should not gain unrestricted access merely because it passed the perimeter.
Separate policy and enforcement
Section titled “Separate policy and enforcement”| Element | Role in NIST zero trust architecture |
|---|---|
| Policy engine | Evaluates policy and relevant information to decide access |
| Policy administrator | Establishes or shuts down the communication path following the decision |
| Policy enforcement point | Enables, monitors, and terminates the subject-to-resource connection |
These are logical components and may be distributed or combined in products.2 Evaluate the actual trust boundary rather than assuming each diagram box requires a separate appliance.
Combine independent defenses
Section titled “Combine independent defenses”Layered protection should address different ways a scenario can fail. Network restrictions, workload authorization, data encryption, and useful audit records can reduce different exposures. Repeating the same check in two components that share one compromised credential provides less independence than it first appears.
Complete mediation means checking each relevant access rather than trusting an earlier user-interface action. Separation of duties can require a different person to approve a sensitive change from the person submitting it. Both ideas need server-side enforcement, including alternative interfaces and automation paths.
Example: approving payroll changes
Section titled “Example: approving payroll changes”Imagine an employee portal with payroll editing. Employees can read their own records. Payroll specialists propose changes, and authorized approvers release them. The application validates ownership and workflow state for every relevant request; hiding an edit button is insufficient.
If the authorization service fails, the design should avoid silently granting payroll edits. Read-only continuity and emergency administration need their own documented decisions. This illustrates a secure default while recognizing that availability and safety requirements must be assessed for each system.
Avoid common misconceptions
Section titled “Avoid common misconceptions”Zero trust is an architectural approach, not proof that a particular product removes all risk. MFA alone does not establish authorization for every resource. Encryption does not stop an already authorized application from misusing plaintext. Likewise, placing a firewall around a service does not replace checks inside it. Evaluate both control effectiveness and the consequences of component failure.
Continue learning
Section titled “Continue learning”See network segmentation for boundaries and cryptography for the services encryption and signatures can provide.
Footnotes
Section titled “Footnotes”-
ISO, ISO/IEC 27002:2022 public overview. Full standard text is not reproduced here. ↩
-
NIST, SP 800-207: Zero Trust Architecture, sections 2.1 and 3. ↩ ↩2
Independent, unofficial study notes. Not endorsed or reviewed by ISC2. CISSP is a registered trademark of ISC2.
Editorial policy & attribution