The Reverse DNS zone is for mapping internet address to hostname. It
is a basic requirement for proper DNS functionality that the Internet
address of every host maps back to the host's authoritative name.
It is important, and cannot be fully explained here, but you should
see the relevant RFCs to research the requirements to properly
implement Reverse DNS. See, RFC 1912.
http://tools.ietf.org/html/rfc1912
for some usage information (Forward-confirmed Reverse DNS)
Any good DNS book should contain relevant details about how reverse
mapping is implemented in the DNS protocol. Zytrax is one of the
well-known online guides http://www.zytrax.com/books/dns/ch3/
The RIRs also have (limited resources)
https://www.arin.net/resources/request/reversedns.html
APNIC's manual is OK and provides instructions for reverse zone
creation, if you ignore APNIC-specific policy details, such as their
delegation objects/delegation templates.
http://www.apnic.net/__data/assets/pdf_file/0009/9792/Reverse-DNS-manual.pdf
This is not necessarily just about sending e-mail, RDNS verification
and use of forward-confirmed RDNS may have fallen into some disuse in
recent years, with SSH replacing RSH and all, and "hosts.equiv"
going away, but, various internet services have been known to reject
connections intentionally (or accidentally) if reverse DNS is not
setup for an IP address of the peer/requestor, or is not present at
all, e.g. many IRC, NNTP, WHOIS, Finger servers.
It is a default behavior of TCP Wrappers (#define PARANOID), commonly
used to protect programs run by inetd on *ix systems.
Authentication/login protocols such as Kerberos also rely on proper RDNS.
When IP addresses are allocated to you by a regional registry such as
ARIN, or if another provider officially re-assigns and delegates
reverse to your IPs, the reverse DNS authority for those IPs becomes
your org's responsibility to either manage (or delegate further
downstream, using NS records, when appropriate).
..
See RFC1035, "Section 3.5. IN-ADDR.ARPA domain " What you supply
to ARIN are hostnames of some DNS servers that ARIN will delegate
your portions of the reverse DNS space to, from the in-addr.arpa
domain.
For example, if you were allocated the IP block 192.0.0.0/23
then these would be delegated to your listed reverse DNS servers
(by the registry), it is entirely dependant on your allocation:
0.0.192.in-addr.arpa
1.0.192.in-addr.arpa
The expectation is that your DNS servers will serve a PTR record in
the proper zone for each IP address you have assigned to a host
e.g. for that example, your first BIND zone might look like
"
0.0.192.in-addr.arpa. IN SOA ns1.example.com. email.addr. ( 2037011800
7200 7200
604800 7200 )
IN NS ns1.example.com.
IN NS ns2.example.com.
1 IN PTR ip192-0-0-1.example.com.
2 IN PTR ip192-0-0-2.example.com.
3 IN PTR ip192-0-0-3.example.com.
"
Assuming "ip192-0-0-1.example.com." resolved to 192.0.0.1, etc, etc...
And you would have a second zone for the 192.0.1.* IPs
EXCEPT.... that is just an example, don't actually use a hostname
like "ip192-0-0-1.example.com." in real life.
[*] Certain overly aggressive blacklists assume that the host must be
a dynamic / dial-up user due to the presence of "192-0-0-1", which
is recognized to be an IP address, so be careful.