A threat model is a structured way to look for security weaknesses before attackers do. Think like a bank robber studying a blueprint: inspect the blind spots, weak locks, and broken alarms before construction begins. Threat modeling applies that mindset to a system’s architecture, identifying vulnerabilities and mitigations during design rather than waiting for a last-minute security test. section 11.6
Review the system diagram with STRIDE, a repeatable framework for considering six threat categories: Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege. For each risk, identify a mitigation rather than assuming the perimeter will hold. section 11.6 section 11.10
Two questions make the review concrete:
- Spoofing: Can someone pretend to be another user or service? A rogue container might pose as an order service to command a database. Identity providers for people and mutual TLS for service-to-service connections help verify identities. section 11.6
- Tampering: Can an attacker alter data in transit or at rest? Someone might change an order’s shipping address in an intercepted request or modify a database file. TLS 1.2 for external traffic and database-level encryption address these risks. section 11.6
Map the boundaries
An attacker might pivot from an HVAC billing portal toward a cash register. Mapping internal boundaries helps reveal how far a breach could spread instead of treating an external integration as an isolated risk. section 11.7.1
Keep the model useful after design. When adding a security solution or third-party integration, record the threat model and chosen mitigations in an Architecture Decision Record. Revisit the record at least quarterly or when a major integration changes, so controls do not quietly become stale. section 11.7.1










