Your failover is a retry loop pretending to be resilience

A team I worked with had a clean SLA: agent responses in under 8 seconds, 99.5% of the time. Then the...