Network & Infrastructure / BGP Policy & Internet Edge
BFD timer sensitivity and false BGP failovers under CPU load
Short BFD intervals appear to improve convergence in a quiet lab but cause repeated route changes during production CPU spikes or congested control queues.
Environment and inputs: BGP speakers and route reflectors; policy syntax, RPKI integration and address-family behavior depend on the implementation. Diagnostic/design example; not tested on production equipment.
Collect the evidence
Record negotiated intervals, detection multipliers, scheduling delay, packet-loss counters and CPU events on both endpoints. Determine whether BFD is offloaded or handled by a shared process.
Explain the result
BFD detects a loss of timely control exchange, not the physical cause. An aggressive detection window can misclassify processing delay as a path failure.
Plan the fix
Choose intervals supported by the hardware and traffic class, then protect required control traffic without over-prioritizing arbitrary workloads. Review independent physical and protocol failure signals.
Check the result and rollback
Exercise the worst expected control-plane load and a real path failure separately. Revert if false positives increase or measured application convergence no longer meets the recovery objective.