Response playbooks¶
Three correlations, three response states. A Playbook 1 alert represents ground preparation with no active attack yet. Playbook 2 indicates a campaign in progress or recently completed. Playbook 3 is an active control-plane incident in which routing trust has already been abused.
Playbook 1: ROA poisoning¶
When this alert fires, RPKI validation state has shifted without a corresponding routing event. The attack has not started; the conditions for it are being arranged.
Immediate actions:
Flag the affected prefix for heightened BMP monitoring. Any subsequent announcement for that prefix warrants escalated scrutiny.
Review recent ROA changes with the responsible registry or RIR. Unexpected creation, modification, or early expiry of a ROA is the primary signal to investigate.
Prepare prefix-specific filters or de-preference rules in advance, before a routing event materialises.
This alert fires rarely when the baseline is correct. If it fires often, the historical baseline or validator configuration is worth reviewing first.
Escalation: any BGP announcement for the affected prefix following a Playbook 1 alert moves the incident to Playbook 2 or Playbook 3.
Playbook 2: Multi-stage BGP attack¶
When this alert fires, the full attack sequence has occurred or is in progress: announcement, validation endorsement, route acceptance, and likely withdrawal. The window for interception is narrow.
Immediate actions:
Apply prefix filtering or de-preference for the affected prefix on border routers. This limits further traffic redirection while the incident is investigated.
Reroute traffic away from affected paths where alternatives exist.
Escalate as a routing security incident, not a misconfiguration.
Follow-on:
Document the full event sequence: AS_PATH, origin AS, announcement and withdrawal timestamps, and which validators endorsed the route.
Audit ROAs for the affected prefix with the responsible registry or RIR.
Review trust anchors for the autonomous system involved.
Playbook 3: RPKI cover hijack¶
When this alert fires, a prefix has been announced under valid RPKI cover and subsequently withdrawn, consistent with tactical rather than operational use. Routing trust has been successfully abused.
Immediate actions:
Treat as a control-plane security incident, not a misconfiguration or operator error.
Apply immediate prefix filtering or de-preference on affected peers.
Audit ROAs and trust anchors for the hijacked prefix.
Follow-on:
Increase monitoring for related prefixes and AS paths. A confirmed cover hijack may be one event in a longer campaign.
Retain all correlated log evidence: BMP events, validator responses, and timing data.
This is late-stage detection. The route was active; any traffic redirection has already occurred. The response value is containment, attribution, and preventing recurrence.
Testing and validation¶
The red-lantern-sim generates deterministic telemetry for all three playbooks. Running a scenario through the simulator and confirming that the corresponding correlation fires is the practical way to verify detection coverage before relying on it in production.
Playbook 3 practice telemetry is in examples/playbook3-practice.json in the simulator repository.
The training variant (playbook3-training.json) adds annotated debug lines and is more useful for
initial analyst familiarisation than for rule validation.