This article details the methodology for performing long-duration (48–72 hour) Availability Zone (AZ) evacuation drills using AWS Application Recovery Controller (ARC) Zonal Shift. It emphasizes shifting from brief failover tests to sustained N-1 operation to identify time-dependent failures, such as misconfigured auto-scaling policies, stale DNS, and operational process gaps.
Key points
Short-duration disaster recovery tests often fail to uncover issues related to long-lived connections, cron jobs, and time-based operational processes.
Running at N-1 capacity for multiple days forces validation of auto-scaling limits and sustained resource pressure in remaining zones.
Operating under an AZ evacuation requires verifying that secondary maintenance tasks, such as certificate rotation and backups, remain functional.
Utilizing ARC Zonal Shift provides a programmatic way to verify application stability and team operational readiness under actual production traffic patterns.