Concept and mechanism
A WAF applies rules to web application traffic. Its effectiveness depends on inspection placement, available data, and paths that actually traverse it. A component forwarding only opaque TCP cannot analyze JSON fields protected by TLS; it needs HTTP visibility at an authorized point. Draw the client, entry, proxy, and origin, including secondary integrations. If the public hostname traverses WAF but origin remains reachable through another path without equivalent protection, a test on the first path does not establish complete coverage. Architecture should explicitly identify every entry and who maintains the controls for each one.
Guided application
WAF also does not replace application authorization. A request with ordinary syntax can attempt to read a portfolio without permission despite valid authentication. The decision needs user, resource, and operation context. A temporary vulnerability rule reduces exposure but does not fix code: retain remediation, validate mitigation, and define review and retirement. When diagnosing a 403, correlate request and time across components before attributing it to WAF. In a fictional exercise, the portal is protected while a direct integration is not; the task is to correct coverage without interrupting legitimate clients.
A block at the public hostname does not prove protection of direct origin access.
Common pitfalls
WAF as authorization; opaque TLS as visible JSON; DNS as isolation; 403 as an identified cause.
Related topics: Order, actions, and overrides · Parsing and inspection limits · Tuning and false positives
Map coverage and retain application and origin controls.
Reference: OWASP WAF overview · DR WAF 2026-09; selected AWS WAF and OWASP CRS operational concepts