Cache for failure, not just speed
Enable stale-if-error and stale-while-revalidate on critical assets so customers keep loading pages during origin issues.
Set per-path TTLs and purge only what is necessary to avoid self-inflicted outages.
Your CDN is part of uptime; treat cache rules like production code.
Measure health at the edge
Run Watch.Dog synthetics from edge regions to see cache hit ratios and latency changes.
Alert on cache-miss spikes or origin failover events before customers feel them.
Keep certs and DNS aligned
Rotate TLS certificates with overlap windows and verify SNI and OCSP stapling from edge locations.
Coordinate DNS TTLs with cache TTLs so failovers propagate without downtime gaps.