RFC 3021 (rfc3021) - Page 2 of 10
Using 31-Bit Prefixes on IPv4 Point-to-Point Links
Alternative Format: Original Text Document
RFC 3021 31-Bit Prefixes on IPv4 Links December 2000
- Deployment of a new Internet Protocol to increase the size of the
address space. One such protocol, IPv6 [RFC 2460], has been through
the IETF process but has yet to see production deployment. Should
it be, deployed, it will still face a many year transition period.
Prior to the availability of a larger address space, it seems prudent
to consider opportunities for making more efficient use of the
existing address space.
One such (small) opportunity is to change the way that point-to-point
links are numbered. One option, which is used today on some parts of
the Internet, is to simply not number point-to-point links between
routers. While this practice may seem, at first, to handily resolve
the problem, it causes a number of problems of its own, including the
inability to consistently manage the unnumbered link or reach a
router through it, difficulty in management and debugging of those
links, and the lack of standardization [RFC 1812].
In current practice, numbered Internet subnets do not use longer than
a 30-bit subnet mask (in most cases), which requires four addresses
per link - two host addresses, one all-zeros network, and one all-
ones broadcast. This is unfortunate for point-to-point links, since
they can only possibly have two identifying endpoints and don't
support the notion of broadcast - any packet which is transmitted by
one end of a link is always received by the other.
A third option is to use host addresses on both ends of a point-to-
point link. This option provides the same address space savings as
using a 31-bit subnet mask, but may only be used in links using PPP
encapsulation [RFC 1332]. The use of host addresses allows for the
assignment of IP addresses belonging to different networks at each
side of the link, causing link and network management not to be
straight forward.
This document is based on the idea that conserving IP addresses on
point-to-point links (using longer than a 30-bit subnet mask) while
maintaining manageability and standard interaction is possible.
Existing documentation [RFC 950] has already hinted at the possible
use of a 1-bit wide host-number field.
The savings in address space resulting from this change is easily
seen--each point-to-point link in a large network would consume two
addresses instead of four. In a network with 500 point-to-point
links, for example, this practice would amount to a savings of 1000
addresses (the equivalent of four class C address spaces).
Retana, et al. Standards Track