Work out failover time from BFD or hello timers plus switchover, and turn it into packet loss and yearly downtime.
Detection normally dominates, which is why BFD exists: it runs a lightweight session at millisecond intervals instead of relying on protocol hellos measured in seconds. Link-down detection is faster still, but only works when the failure is a local physical one. Availability targets are usually written as nines, and this converts a timer setting into that language — 6 events at a second each is well inside five nines, while 6 events at 40 seconds each is not.
Failover Time
Failover = detection (interval × multiplier, or ~10 ms for link-down) + control-plane switchover + FIB update.
Availability impact
downtime per year = failover time × events; availability = 100 − downtime ÷ 31,536,000,000 ms.
Failover = detection (interval × multiplier, or ~10 ms for link-down) + control-plane switchover + FIB update. Detection normally dominates, which is why BFD exists: it runs a lightweight session at millisecond intervals instead of relying on protocol hellos measured in seconds. Link-down detection is faster still, but only works when the failure is a local physical one.
Availability targets are usually written as nines, and this converts a timer setting into that language — 6 events at a second each is well inside five nines, while 6 events at 40 seconds each is not.
This calculator takes 7 inputs: Detection mechanism, Detection interval, Detection multiplier, Switchover time, FIB update time, Packet rate through the path, Failover events per year. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.
Aggressive timers put load on the platform and risk false positives when the control plane is busy — a single missed window tears down a working path. Verify the platform does BFD in hardware or a dedicated process before going below 300 ms, and never tune it on a software-forwarding device.