RFC 7600

Internet Engineering Task Force (IETF)                        R. Despres
Request for Comments: 7600                                     RD-IPtech
Category: Experimental                                     S. Jiang, Ed.
ISSN: 2070-1721                             Huawei Technologies Co., Ltd
                                                                R. Penno
                                                     Cisco Systems, Inc.
                                                                  Y. Lee
                                                                 G. Chen
                                                            China Mobile
                                                                 M. Chen
                                                              BBIX, Inc.
                                                               July 2015

     IPv4 Residual Deployment via IPv6 - A Stateless Solution (4rd)


   This document specifies a stateless solution for service providers to
   progressively deploy IPv6-only network domains while still offering
   IPv4 service to customers.  The solution's distinctive properties are
   that TCP/UDP IPv4 packets are valid TCP/UDP IPv6 packets during
   domain traversal and that IPv4 fragmentation rules are fully
   preserved end to end.  Each customer can be assigned one public IPv4
   address, several public IPv4 addresses, or a shared address with a
   restricted port set.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for examination, experimental implementation, and

   This document defines an Experimental Protocol for the Internet
   community.  This document is a product of the Internet Engineering
   Task Force (IETF).  It represents the consensus of the IETF
   community.  It has received public review and has been approved for
   publication by the Internet Engineering Steering Group (IESG).  Not
   all documents approved by the IESG are a candidate for any level of
   Internet Standard; see Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at

Despres, et al.               Experimental                      [Page 1]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

Copyright Notice

   Copyright (c) 2015 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Despres, et al.               Experimental                      [Page 2]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

Table of Contents

   1. Introduction ....................................................4
   2. Terminology .....................................................5
   3. The 4rd Model ...................................................7
   4. Protocol Specifications .........................................9
      4.1. NAT44 on CE ................................................9
      4.2. Mapping Rules and Other Domain Parameters .................10
      4.3. Reversible Packet Translations at Domain Entries
           and Exits .................................................11
      4.4. Address Mapping from CE IPv6 Prefixes to 4rd IPv4
           Prefixes ..................................................17
      4.5. Address Mapping from 4rd IPv4 Addresses to 4rd
           IPv6 Addresses ............................................19
      4.6. Fragmentation Processing ..................................23
           4.6.1. Fragmentation at Domain Entry ......................23
           4.6.2. Ports of Fragments Addressed to
                  Shared-Address CEs .................................24
           4.6.3. Packet Identifications from Shared-Address CEs .....26
      4.7. TOS and Traffic Class Processing ..........................26
      4.8. Tunnel-Generated ICMPv6 Error Messages ....................27
      4.9. Provisioning 4rd Parameters to CEs ........................27
   5. Security Considerations ........................................30
   6. IANA Considerations ............................................31
   7. Relationship with Previous Works ...............................31
   8. References .....................................................33
      8.1. Normative References ......................................33
      8.2. Informative References ....................................34
   Appendix A. Textual Representation of Mapping Rules ...............37
   Appendix B. Configuring Multiple Mapping Rules ....................37
   Appendix C. Adding Shared IPv4 Addresses to an IPv6 Network .......39
     C.1. With CEs within CPEs .......................................39
     C.2. With Some CEs behind Third-Party Router CPEs  ..............41
   Appendix D. Replacing Dual-Stack Routing with IPv6-Only Routing ...42
   Appendix E. Adding IPv6 and 4rd Service to a Net-10 Network .......43
   Acknowledgements ..................................................44
   Authors' Addresses ................................................44

Despres, et al.               Experimental                      [Page 3]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

1.  Introduction

   For service providers to progressively deploy IPv6-only network
   domains while still offering IPv4 service to customers, the need for
   a stateless solution, i.e., one where no per-customer state is needed
   in IPv4-IPv6 gateway nodes of the provider, has been discussed in
   [Solutions-4v6].  This document specifies one such solution, named
   "4rd" for IPv4 Residual Deployment.  Its distinctive properties are
   that TCP/UDP IPv4 packets are valid TCP/UDP IPv6 packets during
   domain traversal and that IPv4 fragmentation rules are fully
   preserved end to end.

   Using this solution, IPv4 packets are transparently tunneled across
   IPv6 networks (the reverse of IPv6 Rapid Deployment on IPv4
   Infrastructures (6rd) [RFC5969], in which IPv6 packets are
   statelessly tunneled across IPv4 networks).

   While IPv6 headers are too long to be mapped into IPv4 headers (which
   is why 6rd requires encapsulation of full IPv6 packets in IPv4
   packets), IPv4 headers can be reversibly translated into IPv6 headers
   in such a way that, during IPv6 domain traversal, UDP packets having
   checksums and TCP packets are valid IPv6 packets.  IPv6-only
   middleboxes that perform deep packet inspection can operate on them,
   in particular for port inspection and web caches.

   In order to deal with the IPv4 address shortage, customers can be
   assigned shared public IPv4 addresses with statically assigned
   restricted port sets.  As such, it is a particular application of the
   Address plus Port (A+P) approach [RFC6346].

   Deploying 4rd in networks that have enough public IPv4 addresses,
   customer sites can also be assigned full public IPv4 addresses.  4rd
   also supports scenarios where a set of public IPv4 addresses are
   assigned to customer sites.

   The design of 4rd builds on a number of previous proposals made for
   IPv4-via-IPv6 transition technologies (Section 7).

   In some use cases, IPv4-only applications of 4rd-capable customer
   nodes can also work with stateful NAT64s [RFC6146], provided these
   are upgraded to support 4rd tunnels in addition to their IP/ICMP
   translation [RFC6145].  The advantage is then a more complete IPv4
   transparency than with double translation.

   How the 4rd model fits in the Internet architecture is summarized in
   Section 3.  The protocol specifications are detailed in Section 4.
   Sections 5 and 6 deal with security considerations and IANA
   considerations, respectively.  Previous proposals that influenced

Despres, et al.               Experimental                      [Page 4]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   this specification are listed in Section 7.  A few typical 4rd use
   cases are presented in Appendices A, B, C, D, and E.

2.  Terminology

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   document are to be interpreted as described in [RFC2119].

   ISP: Internet Service Provider.  In this document, the service it
        offers can be DSL, fiber-optics, cable, or mobile.  The ISP can
        also be a private-network operator.

   4rd (IPv4 Residual Deployment):  An extension of the IPv4 service
        where public IPv4 addresses can be statically shared among
        several customer sites, each one being assigned an exclusive
        port set.  This service is supported across IPv6-routing

   4rd domain (or Domain):  An ISP-operated IPv6 network across which
        4rd is supported according to the present specification.

   Tunnel packet:  An IPv6 packet that transparently conveys an IPv4
        packet across a 4rd domain.  Its header has enough information
        to reconstitute the IPv4 header at Domain exit.  Its payload is
        the original IPv4 payload.

   CE (Customer Edge):  A customer-side tunnel endpoint.  It can be in a
        node that is a host, a router, or both.

   BR (Border Relay):  An ISP-side tunnel endpoint.  Because its
        operation is stateless (neither per CE nor per session state),
        it can be replicated in as many nodes as needed for scalability.

   4rd IPv6 address:  IPv6 address used as the destination of a Tunnel
        packet sent to a CE or a BR.

   NAT64+:  An ISP NAT64 [RFC6146] that is upgraded to support 4rd
        tunneling when IPv6 addresses it deals with are 4rd IPv6

   4rd IPv4 address:  A public IPv4 address or, in the case of a shared
        public IPv4 address, a public transport address (public IPv4
        address plus port number).

   PSID (Port-Set Identifier):  A flexible-length field that
        algorithmically identifies a port set.

Despres, et al.               Experimental                      [Page 5]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   4rd IPv4 prefix:  A flexible-length prefix that may be a public IPv4
        prefix, a public IPv4 address, or a public IPv4 address followed
        by a PSID.

   Mapping rule:  A set of parameters that are used by BRs and CEs to
        derive 4rd IPv6 addresses from 4rd IPv4 addresses.  Mapping
        rules are also used by each CE to derive a 4rd IPv4 prefix from
        an IPv6 prefix that has been delegated to it.

   EA bits (Embedded Address bits):  Bits that are the same in a 4rd
        IPv4 address and in the 4rd IPv6 address derived from it.

   BR Mapping rule:  The Mapping rule that is applicable to off-domain
        IPv4 addresses (addresses reachable via BRs).  It can also apply
        to some or all CE-assigned IPv4 addresses.

   CE Mapping rule:  A Mapping rule that is applicable only to
        CE-assigned IPv4 addresses (shared or not).

   NAT64+ Mapping rule:  The Mapping rule that is applicable to IPv4
        addresses reachable via a NAT64+.

   CNP (Checksum Neutrality Preserver):  A field of 4rd IPv6 addresses
        that ensures that TCP-like checksums do not change when IPv4
        addresses are replaced with 4rd IPv6 addresses.

   4rd Tag:  A 16-bit tag whose value allows 4rd CEs, BRs, and NAT64+s
        to distinguish 4rd IPv6 addresses from other IPv6 addresses.

Despres, et al.               Experimental                      [Page 6]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

3.  The 4rd Model

                                    4rd Domain
                       |        IPv6 routing         |
                       |  Enforced ingress filtering | +----------
                  ...  |                             | |
                       |                          +------+
        Customer site  |                          |BR(s) |  IPv4
        +------------+ |      BR IPv6 prefix  --> |and/or| Internet
        | dual-stack | |                          |N4T64+|
        |         +--+ |                          +------+
        |         |CE+-+ <-- a CE IPv6 prefix        | |
        |         +--+ |                             | +----------
        |            | |                             |
        +------------+ |     <--IPv4 tunnels-->      +------------
          => Derived   |  (Mesh or hub-and-spoke     |
        4rd IPv4 prefix|         topologies)         |    IPv6
                       |                             |  Internet
                  ...  |                             |
                       |                             +------------
                      <== one or several Mapping rules
                  (e.g., announced to CEs in stateless DHCPv6)

           Figure 1: The 4rd Model in the Internet Architecture

   How the 4rd model fits in the Internet architecture is represented in
   Figure 1.

   A 4rd domain is an IPv6 network that includes one or several 4rd BRs
   or NAT64+s at its border with the public IPv4 Internet and that can
   advertise its IPv4-IPv6 Mapping rule(s) to CEs according to
   Section 4.9.

   BRs of a 4rd Domain are all identical as far as 4rd is concerned.  In
   a 4rd CE, the IPv4 packets that need to reach a BR will be
   transformed (as detailed in Section 4.3) into IPv6 packets that have
   the same anycast IPv6 prefix, which is the 80-bit BR prefix, in their
   destination addresses.  They are then routed to any of the BRs.  The
   80-bit BR IPv6 prefix is an arbitrarily chosen /64 prefix from the
   IPv6 address space of the network operator and appended with 0x0300
   (16-bit 4rd Tag; see R-9 in Section 4.5).

Despres, et al.               Experimental                      [Page 7]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   Using the Mapping rule that applies, each CE derives its 4rd IPv4
   prefix from its delegated IPv6 prefix, or one of them if it has
   several; see Section 4.4 for details.  If the obtained IPv4 prefix
   has more than 32 bits, the assigned IPv4 address is shared among
   several CEs.  Bits beyond the first 32 specify a set of ports whose
   use is reserved for the CE.

   IPv4 traffic is automatically tunneled across the Domain, in either
   mesh topology or hub-and-spoke topology [RFC4925].  By default, IPv4
   traffic between two CEs follows a direct IPv6 route between them
   (mesh topology).  If the ISP configures the hub-and-spoke option,
   each IPv4 packet from one CE to another is routed via a BR.

   During Domain traversal, each tunneled TCP/UDP IPv4 packet looks like
   a valid TCP/UDP IPv6 packet.  Thus, TCP/UDP access control lists that
   apply to IPv6, and possibly some other functions using deep packet
   inspection, also apply to IPv4.

   In order for IPv4 anti-spoofing protection in CEs and BRs to remain
   effective when combined with 4rd tunneling, ingress filtering
   [RFC3704] has to be in effect in IPv6 (see R-12 and Section 5).

   If an ISP wishes to support dynamic IPv4 address sharing in addition
   to or in place of 4rd stateless address sharing, it can do so by
   means of a stateful NAT64.  By upgrading this NAT to add support for
   4rd tunnels, which makes it a NAT64+, CEs that are assigned no static
   IPv4 space can benefit from complete IPv4 transparency between the CE
   and the NAT64.  (Without this NAT64 upgrade, IPv4 traffic is
   translated to IPv6 and back to IPv4, during which time the DF =
   MF = 1 combination for IPv4, as recommended for host fragmentation in
   Section 8 of [RFC4821], is lost.)

   IPv4 packets are kept unchanged by Domain traversal, except that:

   o  The IPv4 Time To Live (TTL), unless it is 1 or 255 at Domain
      entry, decreases during Domain traversal by the number of
      traversed routers.  This is acceptable because it is undetectable
      end to end and also because TTL values that can be used with some
      protocols to test the adjacency of communicating routers are
      preserved [RFC4271] [RFC5082].  The effect on the traceroute
      utility, which uses TTL expiry to discover routers of end-to-end
      paths, is noted in Section 4.3.

Despres, et al.               Experimental                      [Page 8]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   o  IPv4 packets whose lengths are <= 68 octets always have their
      "Don't Fragment" (DF) flags set to 1 at Domain exit even if they
      had DF = 0 at Domain entry.  This is acceptable because these
      packets are too short to be fragmented [RFC791] and so their DF
      bits have no meaning.  Besides, both [RFC1191] and [RFC4821]
      recommend that sources always set DF to 1.

   o  Unless the Tunnel Traffic Class option applies to a Domain
      (Section 4.2), IPv4 packets may have their Type of Service (TOS)
      fields modified after Domain traversal (Section 4.7).

4.  Protocol Specifications

   This section describes detailed 4rd protocol specifications.  They
   are mainly organized by functions.  As a brief summary:

   o  A 4rd CE MUST follow R-1, R-2, R-3, R-4, R-6, R-7, R-8, R-9, R-10,
      R-11, R-12, R-13, R-14, R-16, R-17, R-18, R-19, R-20, R-21, R-22,
      R-23, R-24, R-25, R-26, and R-27.

   o  A 4rd BR MUST follow R-2, R-3, R-4, R-5, R-6, R-9, R-12, R-13,
      R-14, R-15, R-19, R-20, R-21, R-22, and R-24.

4.1.  NAT44 on CE

   R-1:  A CE node that is assigned a shared public IPv4 address MUST
         include a NAT44 [RFC3022].  This NAT44 MUST only use external
         ports that are in the CE-assigned port set.

   NOTE: This specification only concerns IPv4 communication between
   IPv4-capable endpoints.  For communication between IPv4-only
   endpoints and IPv6-only remote endpoints, the "Bump-in-the-Host"
   (BIH) specification [RFC6535] can be used.  It can coexist in a node
   with the CE function, including scenarios where the IPv4-only
   function is a NAT44 [RFC3022].

Despres, et al.               Experimental                      [Page 9]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

4.2.  Mapping Rules and Other Domain Parameters

   R-2:  CEs and BRs MUST be configured with the following Domain

         A.  One or several Mapping rules, each one comprising the

             1.  Rule IPv4 prefix

             2.  EA-bits length

             3.  Rule IPv6 prefix

             4.  Well-Known Ports (WKPs) authorized (OPTIONAL)

         B.  Domain Path MTU (PMTU)

         C.  Hub-and-spoke topology (Yes or No)

         D.  Tunnel Traffic Class (OPTIONAL)

   "Rule IPv4 prefix" is used to find, by a longest match, which Mapping
   rule applies to a 4rd IPv4 address (Section 4.5).  A Mapping rule
   whose Rule IPv4 prefix is longer than /0 is a CE Mapping rule.  BR
   and NAT64+ Mapping rules, which must apply to all off-domain IPv4
   addresses, have /0 as their Rule IPv4 prefixes.

   "EA-bits length" is the number of bits that are common to 4rd IPv4
   addresses and 4rd IPv6 addresses derived from them.  In a CE Mapping
   rule, it is also the number of bits that are common to a CE-delegated
   IPv6 prefix and the 4rd IPv4 prefix derived from it.  BR and NAT64+
   Mapping rules have EA-bits lengths equal to 32.

   "Rule IPv6 prefix" is the prefix that is used as a substitute for the
   Rule IPv4 prefix when a 4rd IPv6 address is derived from a 4rd IPv4
   address (Section 4.5).  In a BR Mapping rule or a NAT64+ Mapping
   rule, it MUST be a /80 prefix whose bits 64-79 are the 4rd Tag.

   "WKPs authorized" may be set for Mapping rules that assign shared
   IPv4 addresses to CEs.  (These rules are those whose length of the
   Rule IPv4 prefix plus the EA-bits length exceeds 32.)  If set,
   well-known ports may be assigned to some CEs having particular IPv6
   prefixes.  If not set, fairness is privileged: all IPv6 prefixes
   concerned with the Mapping rule have port sets having identical
   values (no port set includes any of the well-known ports).

Despres, et al.               Experimental                     [Page 10]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   "Domain PMTU" is the IPv6 Path MTU that the ISP can guarantee for all
   of its IPv6 paths between CEs and between BRs and CEs.  It MUST be at
   least 1280 octets [RFC2460].

   "Hub-and-spoke topology", if set to Yes, requires CEs to tunnel all
   IPv4 packets via BRs.  If set to No, CE-to-CE packets take the same
   routes as native IPv6 packets between the same CEs (mesh topology).

   "Tunnel Traffic Class", if provided, is the IPv6 traffic class that
   BRs and CEs MUST set in Tunnel packets.  In this case, evolutions of
   the IPv6 traffic class that may occur during Domain traversal are not
   reflected in TOS fields of IPv4 packets at Domain exit (Section 4.7).

4.3.  Reversible Packet Translations at Domain Entries and Exits

   R-3:  Domain-entry nodes that receive IPv4 packets with IPv4 options
         MUST discard these packets and return ICMPv4 error messages to
         signal IPv4-option incompatibility (Type = 12, Code = 0,
         Pointer = 20) [RFC792].  This limitation is acceptable because
         there are a lot of firewalls in the current IPv4 Internet that
         also filter IPv4 packets with IPv4 options.

   R-4:  Domain-entry nodes that receive IPv4 packets without IPv4
         options MUST convert them to Tunnel packets, with or without
         IPv6 fragment headers, depending on what is needed to ensure
         IPv4 transparency (Figure 2).  Domain-exit nodes MUST convert
         them back to IPv4 packets.

         An IPv6 fragmentation header MUST be included at tunnel entry
         (Figure 2) if and only if one or several of the following
         conditions hold:

         *  The Tunnel Traffic Class option applies to the Domain.

         *  TTL = 1 OR TTL = 255.

         *  The IPv4 packet is already fragmented, or may be fragmented
            later on, i.e., if MF = 1 OR offset > 0 OR (total length >
            68 AND DF = 0).

         In order to optimize cases where fragmentation headers are
         unnecessary, the NAT44 of a CE that has one SHOULD send packets
         with TTL = 254.

Despres, et al.               Experimental                     [Page 11]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   R-5:  In Domains whose chosen topology is hub-and-spoke, BRs that
         receive 4rd IPv6 packets whose embedded destination IPv4
         addresses match a CE Mapping rule MUST do the equivalent of
         reversibly translating their headers to IPv4 and then
         reversibly translate them back to IPv6 as though packets would
         be entering the Domain.

                     (A) Without IPv6 fragment header
            IPv4 packet                          Tunnel packet
       +--------------------+ :            : +--------------------+
     20|     IPv4 Header    | :    <==>    : |     IPv6 Header    | 40
       +--------------------+ :            : +--------------------+
       |     IP Payload     |      <==>      |     IP Payload     |
       |                    |     Layer 4    |                    |
       +--------------------+    unchanged   +--------------------+

                     (B) With IPv6 fragment header
                                                 Tunnel packet
                                           : +--------------------+
            IPv4 packet                    : |     IPv6 Header    | 40
       +--------------------+ :            : +--------------------+
     20|     IPv4 Header    | :    <==>    : |IPv6 Fragment Header|  8
       +--------------------+ :            : +--------------------+
       |     IP Payload     |      <==>      |     IP Payload     |
       |                    |     Layer 4    |                    |
       +--------------------+    unchanged   +--------------------+

                  Figure 2: Reversible Packet Translation

Despres, et al.               Experimental                     [Page 12]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   R-6:  Values to be set in IPv6 header fields at Domain entry are
         detailed in Table 1 (no fragment header) and Table 2 (with
         fragment header).  Those to be set in IPv4 header fields at
         Domain exit are detailed in Table 3 (no fragment header) and
         Table 4 (with fragment header).

         To convey IPv4 header information that has no equivalent in
         IPv6, some ad hoc fields are placed in IPv6 flow labels and in
         Identification fields of IPv6 fragment headers, as detailed in
         Figure 3.

                    |0      |4                            19|
                    |   0   |         Addr_Prot_Cksm        |
                               IPv6 Flow Label

       0 1 2          |8              |16                           31|
      |.|.|.|    0    |    IPv4_TOS   |             IPv4_ID           |
     /     \ TTL_255         IPv6 Identification Field
   IPv4_DF  TTL_1            (in fragment header if needed)

       Figure 3: 4rd Identification Fields of IPv6 Fragment Headers

     | IPv6 Field          | Value (fields from IPv4 header)        |
     | Version             | 6                                      |
     | Traffic Class       | TOS                                    |
     | Addr_Prot_Cksm      | Sum of addresses and Protocol (Note 1) |
     | Payload length      | Total length - 20                      |
     | Next header         | Protocol                               |
     | Hop limit           | Time to Live                           |
     | Source address      | See Section 4.5                        |
     | Destination address | See Section 4.5                        |

            Table 1: IPv4-to-IPv6 Reversible Header Translation
                         (without Fragment Header)

Despres, et al.               Experimental                     [Page 13]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

    | IPv6 Field      | Value (fields from IPv4 header)              |
    | Version         | 6                                            |
    | Traffic Class   | TOS OR Tunnel Traffic Class (Section 4.7)    |
    | Addr_Prot_Cksm  | Sum of addresses and Protocol (Note 1)       |
    | Payload length  | Total length - 12                            |
    | Next header     | 44 (fragment header)                         |
    | Hop limit       | IF Time to Live = 1 or 255 THEN 254          |
    |                 |   ELSE Time to Live (Note 2)                 |
    | Source address  | See Section 4.5                              |
    | Dest. address   | See Section 4.5                              |
    | 2nd next header | Protocol                                     |
    | Fragment offset | IPv4 fragment offset                         |
    | M               | More Fragments flag (MF)                     |
    | IPv4_DF         | Don't Fragment flag (DF)                     |
    | TTL_1           | IF Time to Live = 1 THEN 1 ELSE 0 (Note 2)   |
    | TTL_255         | IF Time to Live = 255 THEN 1 ELSE 0 (Note 2) |
    | IPv4_TOS        | Type of Service (TOS)                        |
    | IPv4_ID         | Identification                               |

            Table 2: IPv4-to-IPv6 Reversible Header Translation
                          (with Fragment Header)

         | IPv4 Field      | Value (fields from IPv6 header)    |
         | Version         | 4                                  |
         | Header length   | 5                                  |
         | TOS             | Traffic Class                      |
         | Total length    | Payload length + 20                |
         | Identification  | 0                                  |
         | DF              | 1                                  |
         | MF              | 0                                  |
         | Fragment offset | 0                                  |
         | Time to Live    | Hop count                          |
         | Protocol        | Next header                        |
         | Header checksum | Computed as per [RFC791] (Note 3)  |
         | Source address  | Bits 80-111 of source address      |
         | Dest. address   | Bits 80-111 of destination address |

            Table 3: IPv6-to-IPv4 Reversible Header Translation
                         (without Fragment Header)

Despres, et al.               Experimental                     [Page 14]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

    | IPv4 Field            | Value (fields from IPv6 header)         |
    | Version               | 4                                       |
    | Header length         | 5                                       |
    | TOS                   | Traffic Class OR IPv4_TOS (Section 4.7) |
    | Total length          | Payload length + 12                     |
    | Identification        | IPv4_ID                                 |
    | DF                    | IPv4_DF                                 |
    | MF                    | M                                       |
    | Fragment offset       | Fragment offset                         |
    | Time to Live (Note 2) | IF TTL_255 = 1 THEN 255                 |
    |                       |   ELSEIF TTL_1 = 1 THEN 1               |
    |                       |   ELSE hop count                        |
    | Protocol              | 2nd next header                         |
    | Header checksum       | Computed as per [RFC791] (Note 3)       |
    | Source address        | Bits 80-111 of source address           |
    | Destination address   | Bits 80-111 of destination address      |

            Table 4: IPv6-to-IPv4 Reversible Header Translation
                          (with Fragment Header)

   NOTE 1: The need to save in the IPv6 header a checksum of both IPv4
   addresses and the IPv4 protocol field results from the following
   facts: (1) header checksums, present in IPv4 but not in IPv6, protect
   addresses or protocol integrity; (2) in IPv4, ICMP messages and
   null-checksum UDP datagrams depend on this protection because, unlike
   other datagrams, they have no other address-and-protocol integrity
   protection.  The sum MUST be performed in ordinary two's complement

   IP-layer Packet length is another field covered by the IPv4 header
   checksum.  It is not included in the saved checksum because (1) doing
   so would have conflicted with [RFC6437] (flow labels must be the same
   in all packets of each flow); (2) ICMPv4 messages have good enough
   protection with their own checksums; (3) the UDP length field
   provides to null-checksum UDP datagrams the same level of protection
   after Domain traversal as without Domain traversal (consistency
   between IP-layer and UDP-layer lengths can be checked).

Despres, et al.               Experimental                     [Page 15]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   NOTE 2: TTL treatment has been chosen to permit adjacency tests
   between two IPv4 nodes situated at both ends of a 4rd tunnel.  TTL
   values to be preserved for this are TTL = 255 and TTL = 1.  For other
   values, TTL decreases between two IPv4 nodes as though the traversed
   IPv6 routers were IPv4 routers.

   The effect of this TTL treatment on IPv4 traceroute is specific:
   (1) the number of routers of the end-to-end path includes traversed
   IPv6 routers; (2) IPv6 routers of a Domain are listed after IPv4
   routers of Domain entry and exit; (3) the IPv4 address shown for an
   IPv6 router is the IPv6-only dummy IPv4 address (Section 4.8);
   (4) the response time indicated for an IPv6 router is that of the
   next router.

   NOTE 3: Provided the sum of obtained IPv4 addresses and protocol
   matches Addr_Prot_Cksm.  If not, the packet MUST be silently

Despres, et al.               Experimental                     [Page 16]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

4.4.  Address Mapping from CE IPv6 Prefixes to 4rd IPv4 Prefixes

     |             CE IPv6 prefix           |
     :     Longest match        :           :
     :  with a Rule IPv6 prefix :           :
     :           ||             :  EA-bits  :
     :           \/             :   length  :
     +--------------------------+     |     :
     |    Rule IPv6 prefix      |<----'---->:
     +--------------------------+           :
                   ||           :           :
                   \/           :           :
              |Rule IPv4 prefix |  EA bits  |
              :                             :
              |     CE 4rd IPv4 prefix      |
     ________/ \_________                   :
    /                    \                  :
   :                  ____:________________/ \__
   :                 /    :                     \
   :    <= 32       :     :          > 32        :
   +----------------+     +-----------------+----+
   |IPv4 prfx or add|  OR |   IPv4 address  |PSID|
   +----------------+     +-----------------+----+
                          :       32        : || :
                    (by default)          (If WKPs authorized)
                        :    :                     :    :
                    +---+----+---------+           +----+-------------+
      Ports in      |> 0|PSID|any value|    OR     |PSID|  any value  |
   the CE port set  +---+----+---------+           +----+-------------+
                    : 4 :     12       :           :        16        :

      Figure 4: From CE IPv6 Prefix to 4rd IPv4 Address and Port Set

   R-7:  A CE whose delegated IPv6 prefix matches the Rule IPv6 prefix
         of one or several Mapping rules MUST select the CE Mapping rule
         for which the match is the longest.  It then derives its 4rd
         IPv4 prefix as shown in Figure 4: (1) The CE replaces the Rule
         IPv6 prefix with the Rule IPv4 prefix.  The result is the CE
         4rd IPv4 prefix.  (2) If this CE 4rd IPv4 prefix has less than
         32 bits, the CE takes it as its assigned IPv4 prefix.  If it
         has exactly 32 bits, the CE takes it as its IPv4 address.  If

Despres, et al.               Experimental                     [Page 17]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

         it has more than 32 bits, the CE MUST take the first 32 bits as
         its shared public IPv4 address and bits beyond the first 32 as
         its Port-Set identifier (PSID).  Ports of its restricted port
         set are by default those that have any non-zero value in their
         first 4 bits (the PSID offset), followed by the PSID, and
         followed by any values in remaining bits.  If the WKP
         authorized option applies to the Mapping rule, there is no
         4-bit offset before the PSID so that all ports can be assigned.

         NOTE: The choice of the default PSID position in port fields
         has been guided by the following objectives: (1) for fairness,
         avoid having any of the well-known ports 0-1023 in the port set
         specified by any PSID value; (2) for compatibility with RTP/
         RTCP [RFC4961], include in each port set pairs of consecutive
         ports; (3) in order to facilitate operation and training, have
         the PSID at a fixed position in port fields; (4) in order to
         facilitate documentation in hexadecimal notation, and to
         facilitate maintenance, have this position nibble-aligned.
         Ports that are excluded from assignment to CEs are 0-4095,
         instead of just 0-1023, in a trade-off to favor nibble
         alignment of PSIDs and overall simplicity.

   R-8:  A CE whose delegated IPv6 prefix has its longest match with the
         Rule IPv6 prefix of the BR Mapping rule MUST take as its IPv4
         address the 32 bits that, in the delegated IPv6 prefix, follow
         this Rule IPv6 prefix.  If this is the case while the hub-and-
         spoke option applies to the Domain, or if the Rule IPv6 prefix
         is not a /80, there is a configuration error in the Domain.  An
         implementation-dependent administrative action MAY be taken.

         A CE whose delegated IPv6 prefix does not match the Rule IPv6
         prefix of either any CE Mapping rule or the BR Mapping rule,
         and is in a Domain that has a NAT64+ Mapping rule, MUST be
         noted as having the unspecified IPv4 address.

Despres, et al.               Experimental                     [Page 18]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

4.5.  Address Mapping from 4rd IPv4 Addresses to 4rd IPv6 Addresses

   :            32              :  :       16      : \
   +----------------------------+  +---------------+ |
   |         IPv4 address       |  |Port_or_ICMP_ID| |  Shared-address
   +----------------------------+  +---+------+----+ |       case
   :      Longest match         :  : 4 : PSID :      |   (PSID length
   :  with a Rule IPv4 prefix   :      :length:      |  of the rule > 0)
   :       ||                   :      :      :      |    with WKPs
   :       \/                   :      :      :      |  not authorized
   +----------------+-----------+      +------+      | (PSID offset = 4)
   |Rule IPv4 prefix|IPv4 suffix|      | PSID |      |
   +----------------+-----------+      +------+      |
   :       ||        \_______    \____ |      |      |
   :       \/                \        \|      /      |
   +--------------------------+--------+-----+      /
   |    Rule IPv6 prefix      |    EA bits   |
   :                                         :
   |                 IPv6 prefix             |
   :\_______________________________        / \
   :             ___________________\______/   \_______________
   :            /                    \                         \
   :           / (CE Mapping rule)    \   (BR Mapping rule)     \
   :   <= 64  :                        :          112            :
   +----------+---+---+------+---+     +--------------+---+------+---+
   |CE v6 prfx| 0 |tag|v4 add|CNP|     |BR IPv6 prefix|tag|v4 add|CNP|
   +----------+-|-+---+------+---+     +--------------+---+------+---+
   :   <= 64  : | :16 :  32  :16 :     :      64      :16 :  32  :16 :
          Padding to /64

            Figure 5: From 4rd IPv4 Address to 4rd IPv6 Address

Despres, et al.               Experimental                     [Page 19]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   R-9:  BRs, and CEs that are assigned public IPv4 addresses, shared or
         not, MUST derive 4rd IPv6 addresses from 4rd IPv4 addresses via
         the steps below or their functional equivalent (Figure 5
         details the shared public IPv4 address case):

         NOTE: The rules for forming 4rd-specific Interface Identifiers
         (IIDs) are to obey [RFC7136]:

         "Specifications of other forms of 64-bit IIDs MUST specify how
         all 64 bits are set."


         "the whole IID value MUST be viewed as an opaque bit string by
         third parties, except possibly in the local context."

         (1)  If hub-and-spoke topology does not apply to the Domain, or
              if it applies but the IPv6 address to be derived is a
              source address from a CE or a destination address from a
              BR, find the CE Mapping rule whose Rule IPv4 prefix has
              the longest match with the IPv4 address.

              If no Mapping rule is thus obtained, take the BR Mapping

              If the obtained Mapping rule assigns IPv4 prefixes to CEs,
              i.e., if the length of the Rule IPv4 prefix plus EA-bits
              length is 32 - k, with k >= 0, delete the last k bits of
              the IPv4 address.

              Otherwise, if the length of the Rule IPv4 prefix plus the
              EA-bits length is 32 + k, with k > 0, take k as the PSID
              length and append to the IPv4 address the PSID copied from
              bits p to p+3 of the Port_or_ICMP_ID field where (1) p,
              the PSID offset, is 4 by default and 0 if the WKPs
              authorized option applies to the rule; (2) the
              Port_or_ICMP_ID field is in bits of the IP payload that
              depend on whether the address is source or destination, on
              whether the packet is ICMP or not, and, if it is ICMP,
              whether it is an error message or an Echo message.  This
              field is:

              a.  If the packet Protocol is not ICMP, the port field
                  associated with the address (bits 0-15 for a source
                  address and bits 16-31 for a destination address).

              b.  If the packet is an ICMPv4 Echo or Echo reply message,
                  the ICMPv4 Identification field (bits 32-47).

Despres, et al.               Experimental                     [Page 20]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

              c.  If the packet is an ICMPv4 error message, the port
                  field associated with the address in the returned
                  packet header (bits 240-255 for a source address and
                  bits 224-239 for a destination address).

              NOTE 1: Using Identification fields of ICMP messages as
              port fields permits the exchange of Echo requests and Echo
              replies between shared-address CEs and IPv4 hosts having
              exclusive IPv4 addresses.  Echo exchanges between two
              shared-address CEs remain impossible, but this is a
              limitation inherent in address sharing (one reason among
              many to use IPv6).

              NOTE 2: When the PSID is taken in the port fields of the
              IPv4 payload, implementation is kept independent from any
              particular Layer 4 protocol having such port fields by not
              checking that the protocol is indeed one that has such
              port fields.  A packet may consequently go, in the case of
              a source mistake, from a BR to a shared-address CE with a
              protocol that is not supported by this CE.  In this case,
              the CE NAT44 returns an ICMPv4 "protocol unreachable"
              error message.  The IPv4 source is thus appropriately
              informed of its mistake.

         (2)  In the result, replace the Rule IPv4 prefix with the Rule
              IPv6 prefix.

         (3)  If the result is shorter than a /64, append to the result
              a null padding up to 64 bits, followed by the 4rd Tag
              (0x0300), and followed by the IPv4 address.

              NOTE: The 4rd Tag is a 4rd-specific mark.  Its function is
              to ensure that 4rd IPv6 addresses are recognizable by CEs
              without any interference with the choice of subnet
              prefixes in CE sites.  (These choices may have been done
              before 4rd is enabled.)

              For this, the 4rd Tag has its "u" and "g" bits [RFC4291]
              both set to 1, so that they maximally differ from these
              existing IPv6 address schemas.  So far, u = g = 1 has not
              been used in any IPv6 addressing architecture.

              With the 4rd Tag, IPv6 packets can be routed to the 4rd
              function within a CE node based on a /80 prefix that no
              native IPv6 address can contain.

Despres, et al.               Experimental                     [Page 21]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

         (4)  Add to the result a Checksum Neutrality Preserver (CNP).
              Its value, in one's complement arithmetic, is the opposite
              of the sum of 16-bit fields of the IPv6 address other than
              the IPv4 address and the CNP themselves (i.e., five
              consecutive fields in address bits 0-79).

              NOTE: The CNP guarantees that Tunnel packets are valid
              IPv6 packets for all Layer 4 protocols that use the same
              checksum algorithm as TCP.  This guarantee does not depend
              on where the checksum fields of these protocols are placed
              in IP payloads.  (Today, such protocols are UDP [RFC768],
              TCP [RFC793], UDP-Lite [RFC3828], and the Datagram
              Congestion Control Protocol (DCCP) [RFC5595].  Should new
              ones be specified, BRs will support them without needing
              an update.)

   R-10: A 4rd-capable CE SHOULD, and a 4rd-enabled CE MUST, always
         prohibit all addresses that use its advertised prefix and have
         an IID starting with 0x0300 (4rd Tag), by using Duplicate
         Address Detection [RFC4862].

   R-11: A CE that is assigned the unspecified IPv4 address (see
         Section 4.4) MUST use, for packets tunneled between itself and
         the Domain NAT64+, addresses as detailed in Figure 6: part (a)
         for its IPv6 source, and part (b) as IPv6 destinations that
         depend on IPv4 destinations.  A NAT64+, being NAT64 conforming
         [RFC6146], MUST accept IPv6 packets whose destination conforms
         to Figure 6(b) (4rd Tag instead of "u" and 0x00 octets).  In
         its Binding Information Base, it MUST remember whether a
         mapping was created with a "u" or 4rd-tag destination.  In the
         IPv4-to-IPv6 direction, it MUST use 4rd tunneling, with source
         address conforming to Figure 6(b), when using a mapping that
         was created with a 4rd-tag destination.

    (a) |   CE IPv6 prefix    |    0    |4rd Tag|      0      |  CNP |
        :      <= 64          :  >= 0   :   16  :     32      :  16  :
            4rd IPv6 address of a CE having no public IPv4 address

        <----------- Rule IPv6 prefix --------->:
    (b) |      NAT64+ IPv6 prefix       |4rd Tag|IPv4 address |  CNP |
        :               64              :   16  :      32     :  16  :
               4rd IPv6 address of a host reachable via a NAT64+

                     Figure 6: Rules for CE and NAT64+

Despres, et al.               Experimental                     [Page 22]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   R-12: For anti-spoofing protection, CEs and BRs MUST check that the
         IPv6 source address of each received Tunnel packet is that
         which, according to R-9, is derived from the source 4rd IPv4
         address.  For this, the IPv4 address used to obtain the source
         4rd IPv4 address is that embedded in the IPv6 source address
         (in its bits 80-111).  (This verification is needed because
         IPv6 ingress filtering [RFC3704] applies only to IPv6 prefixes,
         without any guarantee that Tunnel packets are built as
         specified in R-9.)

   R-13: For additional protection against packet corruption at a link
         layer that might be undetected at this layer during Domain
         traversal, CEs and BRs SHOULD verify that source and
         destination IPv6 addresses have not been modified.  This can be
         done by checking that they remain checksum neutral (see the
         Note above regarding the CNP).

4.6.  Fragmentation Processing

4.6.1.  Fragmentation at Domain Entry

   R-14: If an IPv4 packet enters a CE or BR with a size such that the
         derived Tunnel packet would be longer than the Domain PMTU, the
         packet has to be either discarded or fragmented.  The
         Domain-entry node MUST discard it if the packet has DF = 1,
         with an ICMP error message returned to the source.  It MUST
         fragment it otherwise, with the payload of each fragment not
         exceeding PMTU - 48.  The first fragment has its offset equal
         to the received offset.  Subsequent fragments have offsets
         increased by the lengths of the payloads of previous fragments.
         Functionally, fragmentation is supposed to be done in IPv4
         before applying reversible header translation to each fragment;
         see Section 4.3.

Despres, et al.               Experimental                     [Page 23]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

4.6.2.  Ports of Fragments Addressed to Shared-Address CEs

   Because ports are available only in the first fragments of IPv4
   fragmented packets, a BR needs a mechanism to send to the right
   shared-address CEs all fragments of fragmented packets.

   For this, a BR MAY systematically reassemble fragmented IPv4 packets
   before tunneling them.  But this consumes large memory space, creates
   opportunities for denial-of-service-attacks, and can significantly
   increase forwarding delays.  This is the reason for the following

   R-15: BRs SHOULD support an algorithm whereby received IPv4 packets
         can be forwarded on the fly.  The following is an example of
         such an algorithm:

         (1)  At BR initialization, if at least one CE Mapping rule
              deals with one or more shared public IPv4 addresses (i.e.,
              length of Rule IPv4 prefix + EA-bits length > 32), the BR
              initializes an empty "IPv4 packet table" whose entries
              have the following items:

                 - IPv4 source

                 - IPv4 destination

                 - IPv4 identification

                 - Destination port

         (2)  When the BR receives an IPv4 packet whose matching Mapping
              rule deals with one or more shared public IPv4 addresses
              (i.e., length of Rule IPv4 prefix + EA-bits length > 32),
              the BR searches the table for an entry whose IPv4 source,
              IPv4 destination, and IPv4 identification are those of the
              received packet.  The BR then performs actions as detailed
              in Table 5, depending on which conditions hold.

Despres, et al.               Experimental                     [Page 24]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

      | - CONDITIONS -              |   |   |   |   |   |   |   |   |
      | First Fragment (offset = 0) | Y | Y | Y | Y | N | N | N | N |
      | Last fragment (MF = 0)      | Y | Y | N | N | Y | Y | N | N |
      | An entry has been found     | Y | N | Y | N | Y | N | Y | N |
      | -------------------------   |   |   |   |   |   |   |   |   |
      | - RESULTING ACTIONS -       |   |   |   |   |   |   |   |   |
      | Create a new entry          | - | - | - | X | - | - | - | - |
      | Use port of the entry       | - | - | - | - | X | - | X | - |
      | Update port of the entry    | - | - | X | - | - | - | - | - |
      | Delete the entry            | X | - | - | - | X | - | - | - |
      | Forward the packet          | X | X | X | X | X | - | X | - |

                            Table 5: BR Actions

         (3)  The BR performs garbage collection for table entries that
              remain unchanged for longer than some limit.  This limit,
              which is normally longer than the maximum time normally
              needed to reassemble a packet, is not critical.  It should
              not, however, be longer than 15 seconds [RFC791].

   R-16: For the above algorithm to be effective, CEs that are assigned
         shared public IPv4 addresses MUST NOT interleave fragments of
         several fragmented packets.

   R-17: CEs that are assigned IPv4 prefixes and are in nodes that route
         public IPv4 addresses rather than only using NAT44s MUST have
         the same behavior as that described just above for BRs.

Despres, et al.               Experimental                     [Page 25]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

4.6.3.  Packet Identifications from Shared-Address CEs

   When packets go from CEs that share the same IPv4 address to a common
   destination, a precaution is needed to guarantee that packet
   identifications set by sources are different.  Otherwise, packet
   reassembly at the destination could be confused because it is based
   only on source IPv4 address and Identification.  The probability of
   such confusing situations may in theory be very low, but a safe
   solution is needed in order to avoid creating new attack

   R-18: A CE that is assigned a shared public IPv4 address MUST only
         use packet identifications that have the CE PSID in their
         bits 0 to PSID length - 1.

   R-19: A BR or a CE that receives a packet from a shared-address CE
         MUST check that bits 0 to PSID length - 1 of their packet
         identifications are equal to the PSID found in the source 4rd
         IPv4 address.

4.7.  TOS and Traffic Class Processing

   IPv4 TOS and IPv6 traffic class have the same semantic, that of the
   differentiated services field, or DS field, specified in [RFC2474]
   and [RFC6040].  Their first 6 bits contain a differentiated services
   codepoint (DSCP), and their last 2 bits can convey explicit
   congestion notifications (ECNs), which both may evolve during Domain
   traversal.  [RFC2983] discusses how the DSCP can be handled by tunnel
   endpoints.  The Tunnel Traffic Class option provides permission to
   ignore DS-field evolutions occurring during Domain traversal, if the
   desired behavior is that of generic tunnels conforming to [RFC2473].

   R-20: Unless the Tunnel Traffic Class option is configured for the
         Domain, BRs and CEs MUST copy the IPv4 TOS into the IPv6
         traffic class at Domain entry and copy back the IPv6 traffic
         class into the IPv4 TOS at Domain exit.

   R-21: If the Tunnel Traffic Class option is configured for a Domain,
         BRs and CEs MUST at Domain entry take the configured Tunnel
         Traffic Class as the IPv6 traffic class and copy the received
         IPv4 TOS into the IPv4_TOS of the fragment header (Figure 3).
         At Domain exit, they MUST copy back the IPv4_TOS of the
         fragment header into the IPv4 TOS.

Despres, et al.               Experimental                     [Page 26]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

4.8.  Tunnel-Generated ICMPv6 Error Messages

   If a Tunnel packet is discarded on its way across a 4rd domain
   because of an unreachable destination, an ICMPv6 error message is
   returned to the IPv6 source.  For the IPv4 source of the discarded
   packet to be informed of packet loss, the ICMPv6 message has to be
   converted into an ICMPv4 message.

   R-22: If a CE or BR receives an ICMPv6 error message [RFC4443], it
         MUST synthesize an ICMPv4 error packet [RFC792].  This packet
         MUST contain the first 8 octets of the discarded packet's IP
         payload.  The reserved IPv4 dummy address (; see
         Section 6) MUST be used as its source address.

         As described in [RFC6145], ICMPv6 Type = 1 and Code = 0
         (Destination Unreachable, No route to destination) MUST be
         translated into ICMPv4 Type = 3 and Code = 0 (Destination
         Unreachable, Net unreachable), and ICMPv6 Type = 3 and Code = 0
         (Time Exceeded, Hop limit exceeded in transit) MUST be
         translated into ICMPv4 Type = 11 and Code = 0 (Time Exceeded,
         time to live exceeded in transit).

4.9.  Provisioning 4rd Parameters to CEs

   Domain parameters listed in Section 4.2 are subject to the following

   R-23: Each Domain MUST have a BR Mapping rule and/or a NAT64+ Mapping
         rule.  The BR Mapping rule is only used by CEs that are
         assigned public IPv4 addresses, shared or not.  The NAT64+
         Mapping rule is only used by CEs that are assigned the
         unspecified IPv4 address (Section 4.4) and therefore need an
         ISP NAT64 to reach IPv4 destinations.

   R-24: Each CE and each BR MUST support up to 32 Mapping rules.

         Support for up to 32 Mapping rules ensures that independently
         acquired CEs and BR nodes can always interwork.

         ISPs that need Mapping rules for more IPv4 prefixes than this
         number SHOULD split their networks into multiple Domains.
         Communication between these domains can be done in IPv4 or by
         some other implementation-dependent, but equivalent, means.

Despres, et al.               Experimental                     [Page 27]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   R-25: For mesh topologies, where CE-CE paths don't go via BRs, all
         Mapping rules of the Domain MUST be sent to all CEs.  For
         hub-and-spoke topologies, where all CE-CE paths go via BRs,
         each CE MAY be sent only the BR Mapping rule of the Domain
         plus, if different, the CE Mapping rule that applies to its CE
         IPv6 prefix.

   R-26: In a Domain where the chosen topology is hub-and-spoke, all CEs
         MUST have IPv6 prefixes that match a CE Mapping rule.
         (Otherwise, packets sent to CEs whose IPv6 prefixes would match
         only the BR Mapping rule would, with longest-match selected
         routes, be routed directly to these CEs.  This would be
         contrary to the hub-and-spoke requirement.)

   R-27: CEs MUST be able to acquire parameters of 4rd domains
         (Section 4.2) in DHCPv6 [RFC3315].  Formats of DHCPv6 options
         to be used are detailed in Figures 7, 8, and 9, with field
         values specified after each figure.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   |      option = OPTION_4RD      |         option-length         |
   |                 encapsulated 4rd rule options                 |

                      Figure 7: DHCPv6 Option for 4rd

   o  option code: 97, OPTION_4RD (see Section 6)

   o  option-length: the length of encapsulated options, in octets

Despres, et al.               Experimental                     [Page 28]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   o  encapsulated 4rd rule options: The OPTION_4RD DHCPv6 option
      contains at least one encapsulated OPTION_4RD_MAP_RULE option and
      a maximum of one encapsulated OPTION_4RD_NON_MAP_RULE option.
      Since DHCP servers normally send whatever options the operator
      configures, operators are advised to configure these options
      appropriately.  DHCP servers MAY check to see that the
      configuration follows these rules and notify the operator in an
      implementation-dependent manner if the settings for these options
      aren't valid.  The length of encapsulated options is in octets.

      0                   1                   2                   3
      0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
     | option = OPTION_4RD_MAP_RULE  |         option-length         |
     |  prefix4-len  |  prefix6-len  |    ea-len     |W|   Reserved  |
     |                    rule-ipv4-prefix                           |
     |                                                               |
     +                                                               +
     |                        rule-ipv6-prefix                       |
     +                                                               +
     |                                                               |

         Figure 8: Encapsulated Option for Mapping-Rule Parameters

   o  option code: 98, encapsulated OPTION_4RD_MAP_RULE option (see
      Section 6)

   o  option-length: 20

   o  prefix4-len: number of bits of the Rule IPv4 prefix

   o  prefix6-len: number of bits of the Rule IPv6 prefix

   o  ea-len: EA-bits length

   o  W: WKP authorized, = 1 if set

   o  rule-ipv4-prefix: Rule IPv4 prefix, left-aligned

   o  rule-ipv6-prefix: Rule IPv6 prefix, left-aligned

Despres, et al.               Experimental                     [Page 29]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

      0                   1                   2                   3
      0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
     |option =OPTION_4RD_NON_MAP_RULE|         option-length         |
     |H|      0    |T| traffic-class |         domain-pmtu           |

       Figure 9: Encapsulated Option for Non-Mapping-Rule Parameters
                              of 4rd Domains

   o  option code: 99, encapsulated OPTION_4RD_NON_MAP_RULE option (see
      Section 6)

   o  option-length: 4

   o  H: Hub-and-spoke topology (= 1 if Yes)

   o  T: Traffic Class flag (= 1 if a Tunnel Traffic Class is provided)

   o  traffic-class: Tunnel Traffic Class

   o  domain-pmtu: Domain PMTU (at least 1280 octets)

   Means other than DHCPv6 that may prove useful to provide 4rd
   parameters to CEs are off-scope for this document.  The same or
   similar parameter formats would, however, be recommended to
   facilitate training and operation.

5.  Security Considerations

   Spoofing attacks

      With IPv6 ingress filtering in effect in the Domain [RFC3704], as
      required in Section 3 (Figure 1 in particular), and with
      consistency checks between 4rd IPv4 and IPv6 addresses
      (Section 4.5), no spoofing opportunity in IPv4 is introduced by
      4rd: being able to use as source IPv6 address only one that has
      been allocated to him, a customer can only provide as source 4rd
      IPv4 address that which derives this IPv6 address according to
      Section 4.5, i.e., one that his ISP has allocated to him.

   Routing loop attacks

      Routing loop attacks that may exist in some "automatic tunneling"
      scenarios are documented in [RFC6324].  No opportunities for
      routing loop attacks have been identified with 4rd.

Despres, et al.               Experimental                     [Page 30]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   Fragmentation-related attacks

      As discussed in Section 4.6, each BR of a Domain that assigns
      shared public IPv4 addresses should maintain a dynamic table of
      fragmented packets that go to these shared-address CEs.

      This leaves BRs vulnerable to denial-of-service attacks from hosts
      that would send very large numbers of first fragments and would
      never send last fragments having the same packet identifications.
      This vulnerability is inherent in IPv4 address sharing, be it
      static or dynamic.  Compared to what it is with algorithms that
      reassemble IPv4 packets in BRs, it is, however, significantly
      mitigated by the algorithm provided in Section 4.6.2, as that
      algorithm uses much less memory space.

6.  IANA Considerations

   IANA has allocated the following:

   o  Encapsulated options OPTION_4RD (97), OPTION_4RD_MAP_RULE (98),
      and OPTION_4RD_NON_MAP_RULE (99).  These options have been
      recorded in the option code space of the "Dynamic Host
      Configuration Protocol for IPv6 (DHCPv6)" registry.  See
      Section 4.9 of this document and Section 24.3 of [RFC3315]).

         Value   |      Description        |  Reference
           97    |       OPTION_4RD        | this document
           98    |   OPTION_4RD_MAP_RULE   | this document
           99    | OPTION_4RD_NON_MAP_RULE | this document

   o  Reserved IPv4 address to be used as the "IPv4 dummy
      address" (Section 4.8).

7.  Relationship with Previous Works

   The present specification has been influenced by many previous IETF
   drafts, in particular those accessible at
   <http://tools.ietf.org/html/draft-xxxx>, where "xxxx" refers to the
   following (listed in order, by date of their first versions):

   o  bagnulo-behave-nat64 (2008-06-10)

   o  xli-behave-ivi (2008-07-06)

   o  despres-sam-scenarios (2008-09-28)

   o  boucadair-port-range (2008-10-23)

Despres, et al.               Experimental                     [Page 31]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   o  ymbk-aplusp (2008-10-27)

   o  xli-behave-divi (2009-10-19)

   o  thaler-port-restricted-ip-issues (2010-02-28)

   o  cui-softwire-host-4over6 (2010-07-06)

   o  dec-stateless-4v6 (2011-03-05)

   o  matsushima-v6ops-transition-experience (2011-03-07)

   o  despres-intarea-4rd (2011-03-07)

   o  deng-aplusp-experiment-results (2011-03-07)

   o  operators-softwire-stateless-4v6-motivation (2011-05-05)

   o  xli-behave-divi-pd (2011-07-04)

   o  murakami-softwire-4rd (2011-07-04)

   o  murakami-softwire-4v6-translation (2011-07-04)

   o  despres-softwire-4rd-addmapping (2011-08-19)

   o  boucadair-softwire-stateless-requirements (2011-09-08)

   o  chen-softwire-4v6-add-format (2011-10-12)

   o  mawatari-softwire-464xlat (2011-10-16)

   o  mdt-softwire-map-dhcp-option (2011-10-24)

   o  mdt-softwire-mapping-address-and-port (2011-10-24)

   o  mdt-softwire-map-translation (2012-01-10)

   o  mdt-softwire-map-encapsulation (2012-01-27)

Despres, et al.               Experimental                     [Page 32]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

8.  References

8.1.  Normative References

   [RFC791]   Postel, J., "Internet Protocol", STD 5, RFC 791,
              DOI 10.17487/RFC0791, September 1981,

   [RFC792]   Postel, J., "Internet Control Message Protocol", STD 5,
              RFC 792, DOI 10.17487/RFC0792, September 1981,

   [RFC793]   Postel, J., "Transmission Control Protocol", STD 7,
              RFC 793, DOI 10.17487/RFC0793, September 1981,

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/
              RFC2119, March 1997,

   [RFC2460]  Deering, S. and R. Hinden, "Internet Protocol, Version 6
              (IPv6) Specification", RFC 2460, DOI 10.17487/RFC2460,
              December 1998,

   [RFC2474]  Nichols, K., Blake, S., Baker, F., and D. Black,
              "Definition of the Differentiated Services Field (DS
              Field) in the IPv4 and IPv6 Headers", RFC 2474,
              DOI 10.17487/RFC2474, December 1998,

   [RFC2983]  Black, D., "Differentiated Services and Tunnels",
              RFC 2983, DOI 10.17487/RFC2983, October 2000,

   [RFC3315]  Droms, R., Ed., Bound, J., Volz, B., Lemon, T., Perkins,
              C., and M. Carney, "Dynamic Host Configuration Protocol
              for IPv6 (DHCPv6)", RFC 3315, DOI 10.17487/RFC3315,
              July 2003,

   [RFC4291]  Hinden, R. and S. Deering, "IP Version 6 Addressing
              Architecture", RFC 4291, DOI 10.17487/RFC4291,
              February 2006,

Despres, et al.               Experimental                     [Page 33]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   [RFC4443]  Conta, A., Deering, S., and M. Gupta, Ed., "Internet
              Control Message Protocol (ICMPv6) for the Internet
              Protocol Version 6 (IPv6) Specification", RFC 4443,
              DOI 10.17487/RFC4443, March 2006,

   [RFC4862]  Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless
              Address Autoconfiguration", RFC 4862,
              DOI 10.17487/RFC4862, September 2007,

   [RFC4925]  Li, X., Ed., Dawkins, S., Ed., Ward, D., Ed., and A.
              Durand, Ed., "Softwire Problem Statement", RFC 4925,
              DOI 10.17487/RFC4925, July 2007,

   [RFC5082]  Gill, V., Heasley, J., Meyer, D., Savola, P., Ed., and C.
              Pignataro, "The Generalized TTL Security Mechanism
              (GTSM)", RFC 5082, DOI 10.17487/RFC5082, October 2007,

   [RFC6040]  Briscoe, B., "Tunnelling of Explicit Congestion
              Notification", RFC 6040, DOI 10.17487/RFC6040,
              November 2010,

8.2.  Informative References

   [NAT444]   Yamagata, I., Shirasaki, Y., Nakagawa, A., Yamaguchi, J.,
              and H. Ashida, "NAT444", Work in Progress,
              draft-shirasaki-nat444-06, July 2012.

   [RFC768]   Postel, J., "User Datagram Protocol", STD 6, RFC 768,
              DOI 10.17487/RFC0768, August 1980,

   [RFC1191]  Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191,
              DOI 10.17487/RFC1191, November 1990,

   [RFC1918]  Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot, G.,
              and E. Lear, "Address Allocation for Private Internets",
              BCP 5, RFC 1918, DOI 10.17487/RFC1918, February 1996,

Despres, et al.               Experimental                     [Page 34]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   [RFC2473]  Conta, A. and S. Deering, "Generic Packet Tunneling in
              IPv6 Specification", RFC 2473, DOI 10.17487/RFC2473,
              December 1998,

   [RFC3022]  Srisuresh, P. and K. Egevang, "Traditional IP Network
              Address Translator (Traditional NAT)", RFC 3022,
              DOI 10.17487/RFC3022, January 2001,

   [RFC3704]  Baker, F. and P. Savola, "Ingress Filtering for Multihomed
              Networks", BCP 84, RFC 3704, DOI 10.17487/RFC3704,
              March 2004,

   [RFC3828]  Larzon, L-A., Degermark, M., Pink, S., Jonsson, L-E., Ed.,
              and G. Fairhurst, Ed., "The Lightweight User Datagram
              Protocol (UDP-Lite)", RFC 3828, DOI 10.17487/RFC3828,
              July 2004,

   [RFC4271]  Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed.,
              "A Border Gateway Protocol 4 (BGP-4)", RFC 4271,
              DOI 10.17487/RFC4271, January 2006,

   [RFC4821]  Mathis, M. and J. Heffner, "Packetization Layer Path MTU
              Discovery", RFC 4821, DOI 10.17487/RFC4821, March 2007,

   [RFC4961]  Wing, D., "Symmetric RTP / RTP Control Protocol (RTCP)",
              BCP 131, RFC 4961, DOI 10.17487/RFC4961, July 2007,

   [RFC5595]  Fairhurst, G., "The Datagram Congestion Control Protocol
              (DCCP) Service Codes", RFC 5595, DOI 10.17487/RFC5595,
              September 2009,

   [RFC5969]  Townsley, W. and O. Troan, "IPv6 Rapid Deployment on IPv4
              Infrastructures (6rd) -- Protocol Specification", RFC
              5969, DOI 10.17487/RFC5969, August 2010,

   [RFC6145]  Li, X., Bao, C., and F. Baker, "IP/ICMP Translation
              Algorithm", RFC 6145, DOI 10.17487/RFC6145, April 2011,

Despres, et al.               Experimental                     [Page 35]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   [RFC6146]  Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful
              NAT64: Network Address and Protocol Translation from IPv6
              Clients to IPv4 Servers", RFC 6146, DOI 10.17487/RFC6146,
              April 2011,

   [RFC6324]  Nakibly, G. and F. Templin, "Routing Loop Attack Using
              IPv6 Automatic Tunnels: Problem Statement and Proposed
              Mitigations", RFC 6324, DOI 10.17487/RFC6324, August 2011,

   [RFC6346]  Bush, R., Ed., "The Address plus Port (A+P) Approach to
              the IPv4 Address Shortage", RFC 6346, DOI 10.17487/
              RFC6346, August 2011,

   [RFC6437]  Amante, S., Carpenter, B., Jiang, S., and J. Rajahalme,
              "IPv6 Flow Label Specification", RFC 6437,
              DOI 10.17487/RFC6437, November 2011,

   [RFC6535]  Huang, B., Deng, H., and T. Savolainen, "Dual-Stack Hosts
              Using "Bump-in-the-Host" (BIH)", RFC 6535,
              DOI 10.17487/RFC6535, February 2012,

   [RFC6887]  Wing, D., Ed., Cheshire, S., Boucadair, M., Penno, R., and
              P. Selkirk, "Port Control Protocol (PCP)", RFC 6887,
              DOI 10.17487/RFC6887, April 2013,

   [RFC7136]  Carpenter, B. and S. Jiang, "Significance of IPv6
              Interface Identifiers", RFC 7136, DOI 10.17487/RFC7136,
              February 2014,

              Boucadair, M., Ed., Matsushima, S., Lee, Y., Bonness, O.,
              Borges, I., and G. Chen, "Motivations for Carrier-side
              Stateless IPv4 over IPv6 Migration Solutions", Work in
              Progress, draft-ietf-softwire-stateless-4v6-motivation-05,
              November 2012.

Despres, et al.               Experimental                     [Page 36]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

Appendix A.  Textual Representation of Mapping Rules

   In the sections that follow, each Mapping rule will be represented as
   follows, using 0bXXX to represent binary number XXX; square brackets
   ("[ ]") indicate optional items:

   {Rule IPv4 prefix, EA-bits length, Rule IPv6 prefix
      [, WKPs authorized]}

    {, 32, 2001:db8:0:1:300::/80}
                               a BR Mapping rule
    {, 22, 2001:db8:4000::/34}
                               a CE Mapping rule
    {, 32, 2001:db8:0:1::/80}
                               a NAT64+ Mapping rule
    {, 22, 2001:db8:4000::/34, Yes}
                               a CE Mapping rule
                                 and hub-and-spoke topology

Appendix B.  Configuring Multiple Mapping Rules

   As far as Mapping rules are concerned, the simplest deployment model
   is that in which the Domain has only one rule (the BR Mapping rule).
   To assign an IPv4 address to a CE in this model, an IPv6 /112 is
   assigned to it, comprising the BR /64 prefix, the 4rd Tag, and the
   IPv4 address.  However, this model has the following limitations: (1)
   shared IPv4 addresses are not supported; (2) IPv6 prefixes used for
   4rd are too long to also be used for native IPv6 addresses; (3) if
   the IPv4 address space of the ISP is split with many disjoint IPv4
   prefixes, the IPv6 routing plan must be as complex as an IPv4 routing
   plan based on these prefixes.

   With more Mapping rules, CE prefixes used for 4rd can be those used
   for native IPv6.  How to choose CE Mapping rules for a particular
   deployment does not need to be standardized.

   The following is only a particular pragmatic approach that can be
   used for various deployment scenarios.  It is applied in some of the
   use cases that follow.

   (1)  Select a "Common_IPv6_prefix" that will appear at the beginning
        of all 4rd CE IPv6 prefixes.

   (2)  Choose all IPv4 prefixes to be used, and assign one of them to
        each CE Mapping rule i.

Despres, et al.               Experimental                     [Page 37]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   (3)  For each CE Mapping rule i, do the following:

        A.  Choose the length of its Rule IPv6 prefix (possibly the same
            for all CE Mapping rules).

        B.  Determine its PSID_length(i).  A CE Mapping rule that
            assigns shared addresses with a sharing ratio of 2^Ki has
            PSID_length = Ki.  A CE Mapping rule that assigns IPv4
            prefixes of length L < 32 is considered to have a negative
            PSID_length (PSID_length = L - 32).

        C.  Derive EA-bits length(i) = 32 - L(Rule IPv4 prefix(i)) +

        D.  Derive the length of Rule_code(i), the prefix to be appended
            to the common prefix to get the Rule IPv6 prefix of rule i:

              L(Rule_code(i)) = L(CE IPv6 prefix(i))
                                - L(Common_IPv6_prefix)
                                - (32 - L(Rule IPv4 prefix(i)))
                                - PSID_length(i)

        E.  Derive Rule_code(i) with the following constraints: (1) its
            length is L(Rule_code(i)); (2) it does not overlap with any
            of the previously obtained Rule_codes (for instance, 010 and
            01011 do overlap, while 00, 011, and 010 do not); (3) it has
            the lowest possible value as a fractional binary number (for
            instance, 0100 < 10 < 11011 < 111).  Thus, rules whose
            Rule_code lengths are 4, 3, 5, and 2 give Rule_codes 0000,
            001, 00010, and 01.

Despres, et al.               Experimental                     [Page 38]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

        F.  Take Rule IPv6 prefix(i) = the Common_IPv6_prefix followed
            by Rule_code(i).

   :<--------------------- L(CE IPv6 prefix(i)) --------------------->:
   :                                                                  :
   :                       32 - L(Rule IPv4 prefix(i))  PSID_length(i):
   :                                                \             |   :
   :                                      :<---------'--------><--'-->:
   :                                      :              ||           :
   :                                      :              \/           :
   :                            :<------->:<--- EA-bits length(i) --->:
   :                          L(Rule_code(i))
   :                            :         :
   |    Common_IPv6_prefix      |Rule_code|
   |                            |   (i)   |
   :<------ L(Rule IPv6 prefix(i)) ------>:

               Figure 10: Diagram of One Pragmatic Approach

Appendix C.  Adding Shared IPv4 Addresses to an IPv6 Network

C.1.  With CEs within CPEs

   Here, we consider an ISP that offers IPv6-only service to up to 2^20
   customers.  Each customer is delegated a /56, starting with common
   prefix 2001:db8:0::/36.  The ISP wants to add public IPv4 service for
   customers that are 4rd capable.  It prefers to do so with stateless
   operation in its nodes but has significantly fewer IPv4 addresses
   than IPv6 addresses, so a sharing ratio is necessary.

   The only IPv4 prefixes it can use are,,, and (neither overlapping nor
   aggregatable).  This gives 2^(32 - 15) + 3 * 2^(32 - 16) IPv4
   addresses, i.e., 2^18 + 2^16.  For the 2^20 customers to have the
   same sharing ratio, the number of IPv4 addresses to be shared has to
   be a power of 2.  The ISP can therefore give up using one of its
   /16s, say the last one.  (Whether or not it could be motivated to
   return it to its Internet Registry is off-scope for this document.)
   The sharing ratio to apply is then 2^20 / 2^18 = 2^2 = 4, giving a
   PSID length of 2.

Despres, et al.               Experimental                     [Page 39]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   Applying the principles of Appendix B with L(Common_IPv6_prefix) =
   36, L(PSID) = 2 for all rules, and L(CE IPv6 prefix(i)) = 56 for all
   rules, Rule_codes and Rule IPv6 prefixes are as follows:

   | CE Rule IPv4 | EA     | Rule-Code | Code      | CE Rule IPv6      |
   | prefix       | bits   | length    | (binary)  | prefix            |
   |              | length |           |           |                   |
   | | 19     | 1         | 0         | 2001:db8:0::/37   |
   | | 18     | 2         | 10        | 2001:db8:800::/38 |
   | | 18     | 2         | 11        | 2001:db8:c00::/38 |

   Mapping rules are then the following:

             {, 19, 2001:0db8:0000::/37}
             {, 18, 2001:0db8:0800::/38}
             {, 18, 2001:0db8:0c00::/38}
             {,    32, 2001:0db8:0000:0001:300::/80}

   The CE whose IPv6 prefix is, for example, 2001:db8:0bbb:bb00::/56
   derives its IPv4 address and its port set as follows (Section 4.4):

      CE IPv6 prefix     : 2001:0db8:0bbb:bb00::/56
      Rule IPv6 prefix(i): 2001:0db8:0800::/38 (longest match)
      EA-bits length(i)  : 18
      EA bits            : 0b11 1011 1011 1011 1011
      Rule IPv4 prefix(i): 0b1100 0000 0000 0100 (
      IPv4 address       : 0b1100 0000 0000 0100 1110 1110 1110 1110
      PSID               : 0b11
      Ports              : 0bYYYY 11XX XXXX XXXX
                           with YYYY > 0, and X...X any value

Despres, et al.               Experimental                     [Page 40]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   An IPv4 packet sent to address and port 7777 is
   tunneled to the IPv6 address obtained as follows (Section 4.5):

      IPv4 address       : (0xc004 eeee)
                         : 0b1100 0000 0000 0100 1110 1110 1110 1110
      Rule IPv4 prefix(i):  (longest match)
                         : 0b1100 0000 0000 0100
      IPv4 suffix(i)     : 0b1110 1110 1110 1110
      EA-bits length(i)  : 18
      PSID length(i)     : 2  (= 16 + 18 - 32)
      Port field         : 0b 0001 1110 0110 0001 (7777)
      PSID               : 0b11
      Rule IPv6 prefix(i): 2001:0db8:0800::/38
      CE IPv6 prefix     : 2001:0db8:0bbb:bb00::/56
      IPv6 address       : 2001:0db8:0bbb:bb00:300:c004:eeee:YYYY
                           with YYYY = the computed CNP

C.2.  With Some CEs behind Third-Party Router CPEs

   We now consider an ISP that has the same need as the ISP described in
   the previous section, except that (1) instead of using only its own
   IPv6 infrastructure, it uses that of a third-party provider and (2)
   some of its customers use this provider's Customer Premises Equipment
   (CPEs) so that they can use specific services offered by the
   provider.  In these CPEs, a non-zero index is used to route IPv6
   packets to the physical port to which CEs are attached, say 0x2.
   Each such CPE delegates to the CE nodes the customer-site IPv6 prefix
   followed by this index.

   The ISP is supposed to have the same IPv4 prefixes as those in the
   previous use case --,, and --
   and to use the same Common_IPv6_prefix, 2001:db8:0::/36.

   We also assume that only a minority of customers use third-party
   CPEs, so that it is sufficient to use only one of the two /16s for

   Mapping rules are then (see Appendix C.1):

             {, 19, 2001:0db8:0000::/37}
             {, 18, 2001:0db8:0800::/38}
             {, 18, 2001:0db8:0c00::/38}
             {,    32, 2001:0db8:0000:0001:300::/80}

   CEs that are behind third-party CPEs derive their own IPv4 addresses
   and port sets as described in Appendix C.1.

Despres, et al.               Experimental                     [Page 41]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   In a BR, and also in a CE if the topology is mesh, the IPv6 address
   that is derived from IPv4 address and port 7777 is
   obtained as described in the previous section, except for the last
   two steps, which are modified as follows:

      IPv4 address       : (0xc004 eeee)
                         : 0b1100 0000 0000 0100 1110 1110 1110 1110
      Rule IPv4 prefix(i):  (longest match)
                         : 0b1100 0000 0000 0100
      IPv4 suffix(i)     : 0b1110 1110 1110 1110
      EA-bits length(i)  : 18
      PSID length(i)     : 2  (= 16 + 18 - 32)
      Port field         : 0b 0001 1110 0110 0001 (7777)
      PSID               : 0b11
      Rule IPv6 prefix(i): 2001:0db8:0800::/38
      CE IPv6 prefix     : 2001:0db8:0bbb:bb00::/60
      IPv6 address       : 2001:0db8:0bbb:bb00:300:
                           with YYYY = the computed CNP

Appendix D.  Replacing Dual-Stack Routing with IPv6-Only Routing

   In this use case, we consider an ISP that offers IPv4 service with
   public addresses individually assigned to its customers.  It also
   offers IPv6 service, as it has deployed dual-stack routing.  Because
   it provides its own CPEs to customers, it can upgrade all of its CPEs
   to support 4rd.  It wishes to take advantage of this capability to
   replace dual-stack routing with IPv6-only routing, without changing
   any IPv4 address or IPv6 prefix.

   For this, the ISP can use the single-rule model described at the
   beginning of Appendix B.  If the prefix routed to BRs is chosen to
   start with 2001:db8:0:1::/64, this rule is:

      {, 32, 2001:db8:0:1:300::/80}

   All that is needed in the network before disabling IPv4 routing is
   the following:

   o  In all routers, where there is an IPv4 route toward x.x.x.x/n, add
      a parallel route toward 2001:db8:0:1:300:x.x.x.x::/(80+n).

   o  Where IPv4 address x.x.x.x was assigned to a CPE, now delegate
      IPv6 prefix 2001:db8:0:1:300:x.x.x.x::/112.

Despres, et al.               Experimental                     [Page 42]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   NOTE: In parallel with this deployment, or after it, shared IPv4
   addresses can be assigned to IPv6 customers.  It is sufficient that
   IPv4 prefixes used for this be different from those used for
   exclusive-address assignments.  Under this constraint, Mapping rules
   can be set up according to the same principles as those described in
   Appendix C.

Appendix E.  Adding IPv6 and 4rd Service to a Net-10 Network

   In this use case, we consider an ISP that has only deployed IPv4,
   possibly because some of its network devices are not yet IPv6
   capable.  Because it did not have enough IPv4 addresses, it has
   assigned private IPv4 addresses [RFC1918] to customers, say 10.x.x.x.
   It thus supports up to 2^24 customers (a "Net-10" network, using the
   NAT444 model [NAT444]).

   Now, it wishes to offer IPv6 service without further delay, using 6rd
   [RFC5969].  It also wishes to offer incoming IPv4 connectivity to its
   customers with a simpler solution than that provided by the Port
   Control Protocol (PCP) [RFC6887].

   This appendix describes an example that adds IPv6 (using 6rd) and 4rd
   services to the "Net-10" private IPv4 network.

   The IPv6 prefix to be used for 6rd is supposed to be 2001:db8::/32,
   and the public IPv4 prefix to be used for shared addresses is
   supposed to be (0xc610).  The resulting sharing ratio
   is 2^24 / 2^(32 - 16) = 256, giving a PSID length of 8.

   The ISP installs one or several BRs at its border to the public IPv4
   Internet.  They support 6rd, and 4rd above it.  The BR prefix /64 is
   supposed to be that which is derived from IPv4 address
   (i.e., 2001:db8:0:100:/64).

   In accordance with [RFC5969], 6rd BRs are configured with the
   following parameters: IPv4MaskLen = 8; 6rdPrefix = 2001:db8::/32;
   6rdBRIPv4Address = (0xc0a80001).

   4rd Mapping rules are then the following:

               {, 24, 2001:db8:0:0:300::/80}
               {,     32, 2001:db8:0:100:300:/80,}

   Any customer device that supports 4rd in addition to 6rd can then use
   its assigned shared IPv4 address with 240 assigned ports.

Despres, et al.               Experimental                     [Page 43]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   If its NAT44 supports port forwarding to provide incoming IPv4
   connectivity (statically, or dynamically with Universal Plug and Play
   (UPnP) and/or the NAT Port Mapping Protocol (NAT-PMP)), it can use it
   with ports of the assigned port set (a possibility that does not
   exist in Net-10 networks without 4rd/6rd).


   This specification has benefited over several years from independent
   proposals, questions, comments, constructive suggestions, and useful
   criticisms from numerous IETF contributors.  The authors would like
   to express recognition of all of these contributors, and especially
   the following, in alphabetical order by their first names: Behcet
   Sarikaya, Bing Liu, Brian Carpenter, Cameron Byrne, Congxiao Bao, Dan
   Wing, Derek Atkins, Erik Kline, Francis Dupont, Gabor Bajko, Hui
   Deng, Jacni Quin (who was an active coauthor of some earlier versions
   of this specification), James Huang, Jan Zorz, Jari Arkko, Kathleen
   Moriarty, Laurent Toutain, Leaf Yeh, Lorenzo Colitti, Marcello
   Bagnulo, Mark Townsley, Mohamed Boucadair, Nejc Skoberne, Olaf
   Maennel, Ole Troan, Olivier Vautrin, Peng Wu, Qiong Sun, Rajiv Asati,
   Ralph Droms, Randy Bush, Satoru Matsushima, Simon Perreault, Stuart
   Cheshire, Suresh Krishnan, Ted Lemon, Teemu Savolainen, Tetsuya
   Murakami, Tina Tsou, Tomek Mrugalski, Washam Fan, Wojciech Dec,
   Xiaohong Deng, Xing Li, and Yu Fu.

Authors' Addresses

   Remi Despres
   3 rue du President Wilson

   Email: despres.remi@laposte.net

   Sheng Jiang (editor)
   Huawei Technologies Co., Ltd
   Q14, Huawei Campus, No. 156 BeiQing Road
   Hai-Dian District, Beijing  100095

   Email: jiangsheng@huawei.com

Despres, et al.               Experimental                     [Page 44]

RFC 7600        Stateless IPv4 Residual Deployment (4rd)       July 2015

   Reinaldo Penno
   Cisco Systems, Inc.
   170 West Tasman Drive
   San Jose, CA  95134
   United States

   Email: repenno@cisco.com

   Yiu Lee
   One Comcast Center
   Philadelphia, PA  19103
   United States

   Email: yiu_lee@cable.comcast.com

   Gang Chen
   China Mobile
   29, Jinrong Avenue
   Xicheng District, Beijing  100033

   Email: phdgang@gmail.com, chengang@chinamobile.com

   Maoke Chen (a.k.a. Noriyuki Arai)
   BBIX, Inc.
   Tokyo Shiodome Building, Higashi-Shimbashi 1-9-1
   Minato-ku, Tokyo  105-7310

   Email: maoke@bbix.net

Despres, et al.               Experimental                     [Page 45]