... it¹

As Albitz and Liu wrote ([AL98]), ``if you're using the Internet, you're already using DNS - even if you don't know it.''

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... it²

At least this is what the specifications state ([Moc87a], [EB97]). As we will see, popular implementations don't always follow the specifications.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... receivers³

It is interesting to denote that this is not always the case with negative caching. In BIND 4.9, for example, the time to live for negatively cached data isn't tunable by the domain administrator and it's hard coded to ten minutes.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... server⁴

Again, a real software implementation detail: some name server implementations don't accept TTLs set to zero, which are changed to 10 minutes.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... present⁵

As a curious data, from the 150000 packets whose TTL is 10 minutes or less (first four figure columns), an approximate 10% corresponds to queries to yahoo.com subdomains.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.