Network Automation, Orchestration, and APIs
Automation layers
Distinguish automation (executing a task reliably), orchestration (coordinating multiple systems toward an outcome), and assurance (verifying the outcome). Know the common interfaces: REST APIs with JSON, NETCONF/YANG, RESTCONF, SNMP for legacy polling, and streaming telemetry for high-frequency state. Understand idempotency, declarative versus imperative models, and why configuration drift matters.
API mechanics
Candidates should be comfortable with HTTP verbs (GET, POST, PUT, PATCH, DELETE), status codes (200/201 success, 400 client error, 401/403 auth, 404 missing, 429 rate limited, 5xx server), authentication methods (tokens, OAuth), and pagination. Know that a PUT typically replaces a resource while PATCH partially updates it. Traps include assuming a 200 response means the change took effect (it may be asynchronous), ignoring rate limits, and hard-coding credentials rather than using secrets management.
Pipelines and safety
Modern designs use version control, templates, and CI/CD pipelines: render configuration from templates, validate in a lab or with dry-run, deploy in stages, and verify with post-change checks. Understand role-based access for automation accounts, audit logging, and the need for rollback. A frequent exam trap is treating automation as inherently safer; without validation and verification, it simply fails faster and at scale.
Worked rule of thumb
For any automation task, apply the loop: intent → render → validate → deploy → verify → record. If a step is missing, the workflow is incomplete. For example, a VLAN change should be templated, validated against the IP plan, pushed to a canary switch, verified by querying the device, and logged with the change ticket.