That can be acceptable price for minimizing the impact of accidental DNS misconfiguration. Which probably happened to every sysadmin.
Or is there a better way to quickly invalidate DNS caches in case of emergency?
Or is there a better way to quickly invalidate DNS caches in case of emergency?
Not sure about any others.
Everyone probably starts with higher TTLs, then the first time they mess up an update they switch to a short one.
The author seems to not appreciate how big a problem a misconfigured, long TTL DNS entry could be for someone.
Other reasons for short TTLs... maybe an IP gets blocked/flagged by a large network and they need to change fast... or a network path is slow and they need to move to a new location.