Question

What is TTL, and why do DNS changes take time?

Vault Verified
Curated Intelligence
Definitive Source
Answer

Time to live is how long a DNS record may be cached before it must be looked up again — and it is the reason a change you made an hour ago is still not visible to everyone.

Why caching exists. Without it, every DNS lookup would go all the way to the authoritative servers, and the system would collapse under load. Caching at every resolver makes DNS fast and survivable. The cost is that the cached copy may be stale.

What TTL controls. Each DNS record carries a TTL in seconds, set by whoever manages the zone. A record with a TTL of 3600 may be held for an hour. Until it expires, every user of that resolver receives the cached answer — regardless of what you changed at the authoritative server.

Why propagation is not really propagation. The term is misleading: nothing is pushed anywhere. Resolvers simply expire their caches at different times, depending on when each first requested the record. Some see the change immediately, some in an hour, some later still, which is why a change appears to roll out gradually.

Why it can take longer than the TTL suggests:

Chained caching. Your device, your router, and your ISP's resolver may all cache independently, each starting its own countdown.

Resolvers that ignore short TTLs, enforcing a minimum of their own.

Browsers cache DNS separately, and frequently for a fixed period regardless of TTL.

Negative caching of a non-existent record, controlled by a separate value in the zone's SOA record — which is why adding a record that someone already tried to look up can take longer to appear than changing an existing one.

How to plan a change properly. This is the practical point: lower the TTL well in advance — to a few minutes — and wait for the old, longer TTL to expire everywhere first. Only then make the change. It will then take effect within minutes. Afterwards, raise the TTL again.

Choosing a TTL. Long TTLs mean better performance, resilience and lower query load. Short TTLs mean faster changes and quicker failover. Records that rarely change deserve long TTLs; records used for failover need short ones.

Related Questions