Routing
Sub-10s SASE Failover
An in-process health daemon switches hub peers before end-user calls can drop.
GA< 10 s
HOW IT WORKS
Walk through a single flow
- 1Probes run on a sub-second cadence with a composite score across ICMP, HTTP, and DNS.
- 2A consecutive-failure threshold triggers the decision to switch.
- 3Peer reassignment and routing changes execute atomically, without cleartext leak or peer overlap.
- 4Hysteresis protects against borderline endpoints causing flap.
UNDER THE HOOD
Technical notes
Why in-process?
The probe loop runs inside the agent rather than on an external schedule, so a target that has gone bad is scored on the next five-second cycle instead of on the next cron run.
RELATED