Failover that has never been drilled
Secondary routing on a diagram is not evidence. Drills and last-incident timelines are.
Continuity documents love dual processors. Operations rooms remember the night the secondary rejected half the MIDs.
Ask for the last drill date, the RTO observed, and the backlog that followed. Absence of those artefacts is a finding.
Manual workarounds need headcount math. A fallback desk of two people cannot cover a regional acceptance spike.
Restore automated settlement after switchover has its own clock. Measure it, or merchants will measure it for you.