Two Ways to Fail Over

A whole region goes dark. You can either flip a DNS failover record or withdraw that region's anycast announcement. Which comparison of the two is accurate?

The withdrawal takes effect as routers reconverge and reaches clients holding a cached answer, while DNS failover cannot beat the TTL plus client-side caches

DNS failover acts sooner because the health check can pull the record the moment the region fails, whereas a withdrawal has to propagate hop by hop across the internet before any client notices

DNS failover is the faster of the two whenever the TTL is set to zero, because a zero TTL obliges every resolver to skip caching and re-query on each lookup

The two are equivalent in speed, since a BGP withdrawal only reaches a client's network once that client's resolver re-queries the authoritative nameserver for the address