MULTIZDB   [plain text]

Sun Jul 24 18:07:09 EDT 1994
This kit consists of the following files:

	MULTIZDB          this file
	MZDB.patch        a patch file
	ns_init.c         complete replacement for nslookup/ns_init.c

(I included a new version of ns_init.c since it was dramatically
rearranged.  There isn't that much new functionality there, but lots
of old things were done in ways inconvenient for this extension.)
ns_init.c and the patches are based on BIND 4.9.3-beta-9.  Sorry in
advance for the difference in code indent style, but somebody can run
it through a beautifier later.

The patch enables MULTIZDB by default, but it has no effect unless you
put "multizone" directives in your named.boot file.

This is a description of the Multizone Database File (MULTIZDB)
extension for BIND.  The overall feature is controlled via an option;
to enable it, define MULTIZDB in "conf/options.h".

MULTIZDB allows you to put records from more than one zone in a single
database file.  For some scenarios, this can simplify database
maintenance, and, by simplifying it, make it less error-prone.  The
MULTIZDB extension affects only the local storage format of the zone
files.  It does not affect any DNS protocol elements nor does it alter
any algorithm by which BIND arrives at an answer.

A simple example is that of a DNS administrator who happens to have
authority for a subdomain and a Class C network address, and where the
two happen to be well-aligned.  The standard way to operate BIND for
this case would be to have two database files.  First, for the forward
zone, e.g., ==>  Second, for the reverse
mapping zone, e.g., ==> db.201.445.23.
Whenever an endpoint is added, deleted, or changed, both database
files must be updated.  With MULTIZDB, both zones could be kept in a
single file, making it more likely that both halves of an update would
be done in synch.

A new keyword, multizone, is recognized for named.boot.  It takes as
its single argument the name of the file containing the resource
records.  Like other directives in named.boot, the filename is
relative to the assumed directory.

A handful of new $ directives are recognized inside a multizone
database file:


These are like the similarly-named directives from the named.boot
file.  For $primary, if a database file is named, it is ignored.
($secondary, $stub, and $cache are provided primarily for completeness
since in most scenarios there is no real win in putting that
information in a multizone database file instead of in named.boot.)
These directives demark zones inside the multizone database file.
Logically, it's as if the reading of a new file were begun at that
point.  Multizone database files do not nest and may not contain
$include directives.
--                    billc@pegasus.ATT.COM    or
  +1 908 576 2932, Fax x4473          William_J_Carpenter@ATT.COM
  AT&T Bell Labs / AT&T EasyLink Services               LZ 3C-207