rfc-3021

Using 31-Bit Prefixes on IPv4 Point-to-Point Links

A. Retana, R. White, V. Fuller, D. McPherson
date2000-12 streamLegacy statusPROPOSED STANDARD pages10 canonicalhttps://www.rfc-editor.org/rfc/rfc3021 doi10.17487/RFC3021
With ever-increasing pressure to conserve IP address space on the Internet, it makes sense to consider where relatively minor changes can be made to fielded practice to improve numbering efficiency. One such change, proposed by this document, is to halve the amount of address space assigned to point-to-point links (common throughout the Internet infrastructure) by allowing the use of 31-bit subnet masks in a very limited way. [STANDARDS-TRACK]

Extracted elements (18)

design-rationale §1

31-bit prefixes are proposed to halve address consumption on point-to-point links from four addresses (two host, one network, one broadcast) to two. A network with 500 point-to-point links saves 1000 addresses — equivalent to four class C spaces.

ip

design-rationale §2.2.1

The loss of directed broadcast functionality on 31-bit prefix links is considered a beneficial side effect rather than a drawback, as it removes a well-known vector for amplification-based DoS attacks (Smurf attacks) on those links.

ip, security

design-rationale §1

Unnumbered point-to-point links were rejected because they prevent consistent management and debugging of the link and lack standardization per RFC 1812. Host-address-only numbering (PPP-only) was rejected because it limits encapsulation choices and complicates link and network management.

ip

interoperability-note §2.3

31-bit prefixes have no impact on current routing protocols. OSPF, IS-IS, BGP, and EIGRP all use multicast, limited broadcast, or unicast for peer communication, none of which are affected by 31-bit subnet masks.

ip, routing

interoperability-note §5

A point-to-point link in which only one end supports 31-bit prefixes may not operate correctly. Both endpoints must implement the behavior defined in this document for the link to function.

ip

interoperability-note §4

Operational testing by at least three ISPs confirmed successful operation with OSPF, IS-IS, BGP, and EIGRP routing protocols when 31-bit prefixes were deployed on point-to-point links.

ip, routing

normative-requirement §2.1 MUST

On a point-to-point link with a 31-bit subnet mask, the two addresses {<Network-number>, 0} and {<Network-number>, -1} MUST be interpreted as host addresses.

ip

normative-requirement §3.1 SHOULD NOT

RFC 1122 new Section 3.2.1.3(h): the subnetwork number {<Network-number>, <Subnet-number>, 0} SHOULD NOT be used as a source address except for endpoints of a 31-bit point-to-point link. For other link types, packets with this destination SHOULD be silently discarded; if not discarded, they MUST be treated as IP broadcasts.

ip

normative-requirement §3.1 MUST NOT

RFC 1122 Section 3.2.1.3(e) update: the directed broadcast address {<Network-number>, <Subnet-number>, -1} MUST NOT be used as a source address, except when the originator is one of the endpoints of a point-to-point link with a 31-bit mask.

ip

normative-requirement §3.3 MUST NOT

RFC 1812 Section 4.2.2.11(d) update: the directed broadcast {<Network-prefix>, -1} MUST NOT be used as a source address except on 31-bit point-to-point link endpoints. A router MAY originate or receive directed broadcasts, but reception MUST be disabled by default.

ip

normative-requirement §3.3 MUST

RFC 1812 Section 4.2.3.1(1) update: in a 31-bit point-to-point link, a packet addressed to {<Network-prefix>, -1} corresponds to one endpoint and MUST be treated as directed to the router on which the address is applied, not as a broadcast.

ip

normative-requirement §3.3 MUST

RFC 1812 Section 4.2.3.1(2) update: in a 31-bit point-to-point link, a packet addressed to {<Network-prefix>, 0} corresponds to one endpoint and MUST be treated as directed to the router on which the address is applied, not as a broadcast.

ip

normative-requirement §3.3 SHOULD

RFC 1812 Section 4.2.3.1(4) update: configuring a 31-bit mask SHOULD allow generation of datagrams addressed to {<Network-prefix>, 0} so that the host-zero address can be used as a valid endpoint address.

ip

normative-requirement §3.3 MUST

The 255.255.255.255 IP broadcast address MUST be used for broadcast Address Mask Replies on point-to-point links with 31-bit subnet masks, since directed broadcast to the link is not possible.

ip

normative-requirement §2.2 MUST

The limited broadcast (255.255.255.255) MUST be used for all broadcast traffic on a point-to-point link with a 31-bit subnet mask, because a directed broadcast to the link is impossible when only two addresses exist.

ip

protocol-element §2.2.1

Directed broadcast to the link subnet is eliminated on 31-bit prefix links: with only two addresses available, there is no distinct all-ones host field to serve as a subnet broadcast destination. Only the limited broadcast {-1,-1} remains available.

ip

protocol-element §2.1

With a 31-bit subnet mask the <Host-number> field is 1 bit wide, yielding exactly two possible addresses: {<Network-number>, 0} and {<Network-number>, -1}. Both addresses MUST be interpreted as host addresses rather than network or broadcast addresses.

ip

security-consideration §6

Using 31-bit subnet masks on core Internet links reduces the attack surface for Smurf-style DoS attacks that rely on directed broadcast packet replication. Eliminating directed broadcasts on these links improves overall network resilience to this class of attack.

ip, security