|
|
BIND, which stands for Berkeley Internet Name Domain, is the most commonly
used implementation of DNS.
DNS is essentially a distributed data base, with control of the different
elements of the data base maintained by individuals responsible for the domain
served by that DNS server. The data is used by DNS servers to assist one host
in identifying the location of another host anywhere in the system,
translating a host name to its IP address, and visa versa.
The DNS distributed data base is much like a directory. It is organized in an
inverted tree fashion, much like the unix directory structure, with the most
inclusive node, or domain, at the top, with multiple levels of sub-domain
names below, until at the end are the actual host names.
Information about each domain, specifying the sub-domains or hosts below it,
are maintained in the DNS data base files. The convention is to call these
files "db files" in BIND 4.X, and "zone files" in BIND 8.x. These files are
made known to the respective DNS server through a configuration file,
named.conf. In earlier versions of BIND, it was called named.boot.
When fully formed, a host name is made up of a sequence of labels separated by
dots. When read from right to left, as DNS parses it, it describes a path
leading from the most inclusive domain in its tree, through successively more
local domains, until its own host name is reached.
Using the full host domain name, this is how a DNS server traverses the DNS
data base, starting at the right-most, most inclusive domain, following data
maintained by the various DNS administrators in their respective data files,
until it finds the target host name, and its IP address.
A domain name is also made up of a sequence of labels separated by dots.
Rather than describing a host, it describes a domain, under which other
sub-domains and/or hosts exist. It can be located in the DNS data base by DNS
servers the same way as was the host domain name.
Sometimes a particular DNS server will not manage an entire domain. Rather,
the domain will be broken up into pieces, called "zones". Responsibility for
these various zones is "delegated" to other DNS servers, and their respective
DNS administrators. So, in DNS configuration files, instead of describing a
domain for which it is responsible, the more general term "zone" is used.
It is also common, in fact recommended, for a DNS Server to have at least one
"backup", another machine that will respond to queries when the main server is
down. The main server is knows as the "master" and the backup as the "slave".
In previous versions of BIND, they were known as "primary" and "secondary".
The rest of this section concerns itself with only "leaf" DNS servers, that
is. servers that only serve hosts. These servers have no domains under it,
only hosts.
There are four types of db or zone files used by a DNS server, each identified
in the server's named.conf file:
- zone.DOMAIN — provides name-to-address mapping
- zone.ADDR — provides address-to-name mapping
- zone.LOCAL — a zone.ADDR file that provides
loopback mapping
- zone.CACHE — a zone.DOMAIN file that identifies
root name servers; also known as the "zone.hint" file.
|