Caching and Time to Live
When a server is processing a recursive query, it might be required to send out several queries to find the definitive answer. The server caches all of the information that it receives during this process for a time that is specified in the returned data. This amount of time is referred to as the time to live (TTL) and is specified in seconds. The server administrator of the primary zone that contains the data decides on the TTL for the data. Smaller TTL values help ensure that information about the domain is more consistent across the network, in the event that this data changes often. However, this also increases the load on the name servers that contain the name, and it also increases Internet traffic. Because data is cached, changes made in resource records might not be immediately available to the entire Internet.
Once data is cached by a DNS server, the DNS server must start decreasing the TTL from its original value so that it will know when to flush the data from its cache. When the DNS server answers a query with its cached data, it includes the remaining TTL for the data. The resolver can then cache this data, using the TTL sent by the server.