Failover RTO Reality: Why Standard Front Door Has a 5-Minute Floor
If your DR plan says "sub-minute RTO" and you're on Front Door Standard, the math doesn't add up. Here's why — and what to pay for if you need better.
Three things contribute to Front Door's failover time, and only one of them is software. The other two are physics and behavior.
1. Probe cadence (the floor)
- Standard SKU: minimum probe interval is 100 seconds.
- Premium SKU: minimum is 30 seconds.
- Origins are marked unhealthy after 3 consecutive failures.
Math: 3 × 100 s = 5 minutes worst-case detection time on Standard. You cannot beat that without paying for Premium.
2. Edge POP propagation
Front Door has 190+ POPs globally. They don't all update routing tables in the same second. There's a propagation delay typically measured in tens of seconds after detection.
3. Client connection pinning
Already-open TCP/TLS sessions stay pinned to the original origin until they close. A user holding a chat-style long-poll connection might keep hitting the failed origin for a minute or more after Front Door has shifted new traffic away.
The honest design rule
If your business RTO is <1 minute, Front Door Standard alone won't get you there. Options:
- Pay for Front Door Premium (30 s probes, ~3× detection speed).
- Add active-active instead of active-passive (no failover; both regions always serve).
- Use Traffic Manager with even shorter probe intervals for non-HTTPS workloads.
And whatever you choose: drill it. Three drills beat any architecture diagram.