Pick the right TTLs
Shorten TTLs for critical endpoints so failover is responsive.
Keep longer TTLs for static assets to reduce query load.
Health check routing
Use DNS providers with integrated health checks and regional probes.
Alert when checks flap so you can tune thresholds.
DNS checklist
- Failover records per region
- Separate health check monitors
- Runbooks for record freezes
Test often
Simulate failover during business hours with stakeholders watching.
Log every DNS change with ticket IDs for traceability.
DNS changes should be rehearsed before production incidents.