RFC 7597

Internet Engineering Task Force (IETF)                     O. Troan, Ed.
Request for Comments: 7597                                        W. Dec
Category: Standards Track                                  Cisco Systems
ISSN: 2070-1721                                                    X. Li
                                                                  C. Bao
                                                     Tsinghua University
                                                           S. Matsushima
                                                        SoftBank Telecom
                                                             T. Murakami
                                                             IP Infusion
                                                          T. Taylor, Ed.
                                                     Huawei Technologies
                                                               July 2015

         Mapping of Address and Port with Encapsulation (MAP-E)


   This document describes a mechanism for transporting IPv4 packets
   across an IPv6 network using IP encapsulation.  It also describes a
   generic mechanism for mapping between IPv6 addresses and IPv4
   addresses as well as transport-layer ports.

Status of This Memo

   This is an Internet Standards Track document.

   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).  Further information on
   Internet Standards is available in 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

Troan, et al.                Standards Track                    [Page 1]

RFC 7597                          MAP-E                        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.

Troan, et al.                Standards Track                    [Page 2]

RFC 7597                          MAP-E                        July 2015

Table of Contents

   1. Introduction ....................................................4
   2. Conventions .....................................................5
   3. Terminology .....................................................5
   4. Architecture ....................................................7
   5. Mapping Algorithm ...............................................8
      5.1. Port-Mapping Algorithm ....................................10
      5.2. Basic Mapping Rule (BMR) ..................................11
      5.3. Forwarding Mapping Rule (FMR) .............................14
      5.4. Destinations outside the MAP Domain .......................14
   6. The IPv6 Interface Identifier ..................................15
   7. MAP Configuration ..............................................15
      7.1. MAP CE ....................................................15
      7.2. MAP BR ....................................................16
   8. Forwarding Considerations ......................................17
      8.1. Receiving Rules ...........................................17
      8.2. ICMP ......................................................18
      8.3. Fragmentation and Path MTU Discovery ......................18
           8.3.1. Fragmentation in the MAP Domain ....................18
           8.3.2. Receiving IPv4 Fragments on the MAP Domain
                  Borders ............................................19
           8.3.3. Sending IPv4 Fragments to the Outside ..............19
   9. NAT44 Considerations ...........................................19
   10. Security Considerations .......................................20
   11. References ....................................................21
      11.1. Normative References .....................................21
      11.2. Informative References ...................................21
   Appendix A. Examples ..............................................25
   Appendix B. A More Detailed Description of the Derivation of the
               Port-Mapping Algorithm ................................29
     B.1. Bit Representation of the Algorithm ........................31
     B.2. GMA Examples ...............................................32
   Acknowledgements ..................................................32
   Contributors ......................................................33
   Authors' Addresses ................................................34

Troan, et al.                Standards Track                    [Page 3]

RFC 7597                          MAP-E                        July 2015

1.  Introduction

   Mapping of IPv4 addresses in IPv6 addresses has been described in
   numerous mechanisms dating back to the mid-1990s [RFC1933] [RFC4213].
   The "automatic tunneling" mechanism as first described in [RFC1933]
   assigned a globally unique IPv6 address to a host by combining the
   host's IPv4 address with a well-known IPv6 prefix.  Given an IPv6
   packet with a destination address with an embedded IPv4 address, a
   node could automatically tunnel this packet by extracting the IPv4
   tunnel endpoint address from the IPv6 destination address.

   There are numerous variations of this idea, as described in 6over4
   [RFC2529], 6to4 [RFC3056], the Intra-Site Automatic Tunnel Addressing
   Protocol (ISATAP) [RFC5214], and IPv6 Rapid Deployment on IPv4
   Infrastructures (6rd) [RFC5969].

   The commonalities of all of these IPv6-over-IPv4 mechanisms are as

   o  Automatic provisioning of an IPv6 address for a host or an IPv6
      prefix for a site.

   o  Algorithmic or implicit address resolution of tunnel endpoint
      addresses.  Given an IPv6 destination address, an IPv4 tunnel
      endpoint address can be calculated.

   o  Embedding of an IPv4 address or part thereof into an IPv6 address.

   In later phases of IPv4-to-IPv6 migration, it is expected that
   IPv6-only networks will be common, while there will still be a need
   for residual IPv4 deployment.  This document describes a generic
   mapping of IPv4 to IPv6 and a mechanism for encapsulating IPv4
   over IPv6.

   Just as for the IPv6-over-IPv4 mechanisms referred to above, the
   residual IPv4-over-IPv6 mechanism must be capable of:

   o  Provisioning an IPv4 prefix, an IPv4 address, or a shared IPv4

   o  Algorithmically mapping between an IPv4 prefix, an IPv4 address,
      or a shared IPv4 address and an IPv6 address.

   The mapping scheme described here supports encapsulation of IPv4
   packets in IPv6 in both mesh and hub-and-spoke topologies, including
   address mappings with full independence between IPv6 and IPv4

Troan, et al.                Standards Track                    [Page 4]

RFC 7597                          MAP-E                        July 2015

   This document describes the delivery of IPv4 unicast service across
   an IPv6 infrastructure.  IPv4 multicast is not considered in this

   The Address plus Port (A+P) architecture of sharing an IPv4 address
   by distributing the port space is described in [RFC6346].
   Specifically, Section 4 of [RFC6346] covers stateless mapping.  The
   corresponding stateful solution, Dual-Stack Lite (DS-Lite), is
   described in [RFC6333].  The motivations for this work are described
   in [Solutions-4v6].

   [RFC7598] defines DHCPv6 options for the provisioning of MAP.  Other
   means of provisioning are possible.  Deployment considerations are
   described in [MAP-Deploy].

   MAP relies on IPv6 and is designed to deliver dual-stack service
   while allowing IPv4 to be phased out within the service provider's
   (SP's) network.  The phasing out of IPv4 within the SP network is
   independent of whether the end user disables IPv4 service or not.
   Further, "greenfield" IPv6-only networks may use MAP in order to
   deliver IPv4 to sites via the IPv6 network.

2.  Conventions

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   document are to be interpreted as described in RFC 2119 [RFC2119].

3.  Terminology

   MAP domain:             One or more MAP Customer Edge (CE) devices
                           and Border Relays (BRs) connected to the same
                           virtual link.  A service provider may deploy
                           a single MAP domain or may utilize multiple
                           MAP domains.

   MAP Rule:               A set of parameters describing the mapping
                           between an IPv4 prefix, IPv4 address, or
                           shared IPv4 address and an IPv6 prefix or
                           address.  Each domain uses a different
                           mapping rule set.

   MAP node:               A device that implements MAP.

Troan, et al.                Standards Track                    [Page 5]

RFC 7597                          MAP-E                        July 2015

   MAP Border Relay (BR):  A MAP-enabled router managed by the service
                           provider at the edge of a MAP domain.  A BR
                           has at least an IPv6-enabled interface and an
                           IPv4 interface connected to the native IPv4
                           network.  A MAP BR may also be referred to as
                           simply a "BR" within the context of MAP.

   MAP Customer Edge (CE): A device functioning as a Customer Edge
                           router in a MAP deployment.  A typical MAP CE
                           adopting MAP Rules will serve a residential
                           site with one WAN-side interface and one or
                           more LAN-side interfaces.  A MAP CE may also
                           be referred to as simply a "CE" within the
                           context of MAP.

   Port set:               Each node has a separate part of the
                           transport-layer port space; this is denoted
                           as a port set.

   Port Set ID (PSID):     Algorithmically identifies a set of ports
                           exclusively assigned to a CE.

   Shared IPv4 address:    An IPv4 address that is shared among multiple
                           CEs.  Only ports that belong to the assigned
                           port set can be used for communication.  Also
                           known as a port-restricted IPv4 address.

   End-user IPv6 prefix:   The IPv6 prefix assigned to an End-user CE by
                           means other than MAP itself, e.g.,
                           provisioned using DHCPv6 Prefix Delegation
                           (PD) [RFC3633], assigned via Stateless
                           Address Autoconfiguration (SLAAC) [RFC4862],
                           or configured manually.  It is unique for
                           each CE.

   MAP IPv6 address:       The IPv6 address used to reach the MAP
                           function of a CE from other CEs and from BRs.

   Rule IPv6 prefix:       An IPv6 prefix assigned by a service provider
                           for a mapping rule.

   Rule IPv4 prefix:       An IPv4 prefix assigned by a service provider
                           for a mapping rule.

Troan, et al.                Standards Track                    [Page 6]

RFC 7597                          MAP-E                        July 2015

   Embedded Address (EA) bits:
                           The IPv4 EA-bits in the IPv6 address identify
                           an IPv4 prefix/address (or part thereof) or a
                           shared IPv4 address (or part thereof) and a
                           Port Set Identifier.

4.  Architecture

   In accordance with the requirements stated above, the MAP mechanism
   can operate with shared IPv4 addresses, full IPv4 addresses, or IPv4
   prefixes.  Operation with shared IPv4 addresses is described here,
   and the differences for full IPv4 addresses and prefixes are
   described below.

   The MAP mechanism uses existing standard building blocks.  The
   existing Network Address and Port Translator (NAPT) [RFC2663] on the
   CE is used with additional support for restricting transport-protocol
   ports, ICMP identifiers, and fragment identifiers to the configured
   port set.  For packets outbound from the private IPv4 network, the CE
   NAPT MUST translate transport identifiers (e.g., TCP and UDP port
   numbers) so that they fall within the CE's assigned port range.

   The NAPT MUST in turn be connected to a MAP-aware forwarding function
   that does encapsulation/decapsulation of IPv4 packets in IPv6.  MAP
   supports the encapsulation mode specified in [RFC2473].  In addition,
   MAP specifies an algorithm to do "address resolution" from an IPv4
   address and port to an IPv6 address.  This algorithmic mapping is
   specified in Section 5.

   The MAP architecture described here restricts the use of the shared
   IPv4 address to only be used as the global address (outside) of the
   NAPT running on the CE.  A shared IPv4 address MUST NOT be used to
   identify an interface.  While it is theoretically possible to make
   host stacks and applications port-aware, it would be a drastic change
   to the IP model [RFC6250].

   For full IPv4 addresses and IPv4 prefixes, the architecture just
   described applies, with two differences: first, a full IPv4 address
   or IPv4 prefix can be used as it is today, e.g., for identifying an
   interface or as a DHCP pool, respectively.  Second, the NAPT is not
   required to restrict the ports used on outgoing packets.

Troan, et al.                Standards Track                    [Page 7]

RFC 7597                          MAP-E                        July 2015

   This architecture is illustrated in Figure 1.

         User N
       Private IPv4
      |  Network
   |  |  MAP CE       |
   | +-----+--------+ |
   | NAPT44|  MAP   | |
   | +-----+        | |\     ,-------.                      .------.
   |       +--------+ | \ ,-'         `-.                 ,-'       `-.
   O------------------O  /              \   O---------O  /   Public   \
                        /    IPv6-only  \  |  MAP    | /     IPv4      \
                       (    Network      --+  Border +-     Network    )
                        \  (MAP Domain) /  |  Relay  | \               /
   O------------------O  \              /   O---------O  \            /
   |    MAP   CE      |  /".         ,-'                 `-.       ,-'
   | +-----+--------+ | /   `----+--'                       ------'
   | NAPT44|  MAP   | |/
   | +-----+        | |
   |   |   +--------+ |
        User M
      Private IPv4

                        Figure 1: Network Topology

   The MAP BR connects one or more MAP domains to external IPv4

5.  Mapping Algorithm

   A MAP node is provisioned with one or more mapping rules.

   Mapping rules are used differently, depending on their function.
   Every MAP node must be provisioned with a Basic Mapping Rule.  This
   is used by the node to configure its IPv4 address, IPv4 prefix, or
   shared IPv4 address.  This same basic rule can also be used for
   forwarding, where an IPv4 destination address and, optionally, a
   destination port are mapped into an IPv6 address.  Additional mapping
   rules are specified to allow for multiple different IPv4 subnets to
   exist within the domain and optimize forwarding between them.

Troan, et al.                Standards Track                    [Page 8]

RFC 7597                          MAP-E                        July 2015

   Traffic outside of the domain (i.e., when the destination IPv4
   address does not match (using longest matching prefix) any Rule IPv4
   prefix in the Rules database) is forwarded to the BR.

   There are two types of mapping rules:

   1.  Basic Mapping Rule (BMR) - mandatory.  A CE can be provisioned
       with multiple End-user IPv6 prefixes.  There can only be one
       Basic Mapping Rule per End-user IPv6 prefix.  However, all CEs
       having End-user IPv6 prefixes within (aggregated by) the same
       Rule IPv6 prefix may share the same Basic Mapping Rule.  In
       combination with the End-user IPv6 prefix, the Basic Mapping Rule
       is used to derive the IPv4 prefix, address, or shared address and
       the PSID assigned to the CE.

   2.  Forwarding Mapping Rule (FMR) - optional; used for forwarding.
       The Basic Mapping Rule may also be a Forwarding Mapping Rule.
       Each Forwarding Mapping Rule will result in an entry in the rule
       table for the Rule IPv4 prefix.  Given a destination IPv4 address
       and port within the MAP domain, a MAP node can use the matching
       FMR to derive the End-user IPv6 address of the interface through
       which that IPv4 destination address and port combination can be
       reached.  In hub-and-spoke mode, there are no FMRs.

   Both mapping rules share the same parameters:

   o  Rule IPv6 prefix (including prefix length)

   o  Rule IPv4 prefix (including prefix length)

   o  Rule EA-bit length (in bits)

   A MAP node finds its BMR by doing a longest match between the
   End-user IPv6 prefix and the Rule IPv6 prefix in the Mapping Rules
   table.  The rule is then used for IPv4 prefix, address, or shared
   address assignment.

   A MAP IPv6 address is formed from the BMR Rule IPv6 prefix.  This
   address MUST be assigned to an interface of the MAP node and is used
   to terminate all MAP traffic being sent or received to the node.

   Port-restricted IPv4 routes are installed in the rule table for all
   the Forwarding Mapping Rules, and a default route is installed to the
   MAP BR (see Section 5.4).

Troan, et al.                Standards Track                    [Page 9]

RFC 7597                          MAP-E                        July 2015

   Forwarding Mapping Rules are used to allow direct communication
   between MAP CEs; this is known as "Mesh mode".  In hub-and-spoke
   mode, there are no Forwarding Mapping Rules; all traffic MUST be
   forwarded directly to the BR.

   While an FMR is optional in the sense that a MAP CE MAY be configured
   with zero or more FMRs -- depending on the deployment -- all MAP CEs
   MUST implement support for both rule types.

5.1.  Port-Mapping Algorithm

   The port-mapping algorithm is used in domains whose rules allow IPv4
   address sharing.

   The simplest way to represent a port range is using a notation
   similar to Classless Inter-Domain Routing (CIDR) [RFC4632].  For
   example, the first 256 ports are represented as port prefix 0.0/8 and
   the last 256 ports as 255.0/8.  In hexadecimal, these would be
   0x0000/8 (PSID = 0) and 0xFF00/8 (PSID = 0xFF), respectively.  Using
   this technique but wishing to avoid allocating the system ports
   [RFC6335] to the user, one would have to exclude the use of one or
   more PSIDs (e.g., PSIDs 0 to 3 in the example just given).

   When the PSID is embedded in the End-user IPv6 prefix, it is
   desirable to minimize the restrictions of possible PSID values in
   order to minimize dependencies between the End-user IPv6 prefix and
   the assigned port set.  This is achieved by using an infix
   representation of the port value.  Using such a representation, the
   well-known ports are excluded by restrictions on the value of the
   high-order bit field (A) rather than the PSID.

   The infix algorithm allocates ports to a given CE as a series of
   contiguous ranges spaced at regular intervals throughout the complete
   range of possible port-set values.

                              0                   1
                              0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
               Ports in      |     A     |    PSID   |   j   |
            the CE port set  |    > 0    |           |       |
                             |  a bits   |  k bits   |m bits |

            Figure 2: Structure of a Port-Restricted Port Field

Troan, et al.                Standards Track                   [Page 10]

RFC 7597                          MAP-E                        July 2015

   a bits:  The number of offset bits -- 6 by default, as this excludes
            the system ports (0-1023).  To guarantee non-overlapping
            port sets, the offset 'a' MUST be the same for every MAP CE
            sharing the same address.

        A:  Selects the range of the port number.  For 'a' > 0, A MUST
            be larger than 0.  This ensures that the algorithm excludes
            the system ports.  For the default value of 'a' (6), the
            system ports are excluded by requiring that A be greater
            than 0.  Smaller values of 'a' exclude a larger initial
            range, e.g., 'a' = 4 will exclude ports 0-4095.  The
            interval between initial port numbers of successive
            contiguous ranges assigned to the same user is 2^(16 - a).

   k bits:  The length in bits of the PSID field.  To guarantee
            non-overlapping port sets, the length 'k' MUST be the same
            for every MAP CE sharing the same address.  The sharing
            ratio is 2^k.  The number of ports assigned to the user is
            2^(16 - k) - 2^m (excluded ports).

     PSID:  The Port Set Identifier (PSID).  Different PSID values
            guarantee non-overlapping port sets, thanks to the
            restrictions on 'a' and 'k' stated above, because the PSID
            always occupies the same bit positions in the port number.

   m bits:  The number of contiguous ports is given by 2^m.

        j:  Selects the specific port within a particular range
            specified by the concatenation of A and the PSID.

5.2.  Basic Mapping Rule (BMR)

   The Basic Mapping Rule is mandatory and is used by the CE to
   provision itself with an IPv4 prefix, IPv4 address, or shared IPv4
   address.  Recall from Section 5 that the BMR consists of the
   following parameters:

   o  Rule IPv6 prefix (including prefix length)

   o  Rule IPv4 prefix (including prefix length)

   o  Rule EA-bit length (in bits)

Troan, et al.                Standards Track                   [Page 11]

RFC 7597                          MAP-E                        July 2015

   Figure 3 shows the structure of the complete MAP IPv6 address as
   specified in this document.

   |     n bits         |  o bits   | s bits  |   128-n-o-s bits      |
   |  Rule IPv6 prefix  |  EA bits  |subnet ID|     interface ID      |
   |<---  End-user IPv6 prefix  --->|

                     Figure 3: MAP IPv6 Address Format

   The Rule IPv6 prefix is common among all CEs using the same Basic
   Mapping Rule within the MAP domain.  The EA bit field encodes the
   CE-specific IPv4 address and port information.  The EA bit field,
   which is unique for a given Rule IPv6 prefix, can contain a full or
   partial IPv4 address and, in the shared IPv4 address case, a PSID.
   An EA bit field length of 0 signifies that all relevant MAP IPv4
   addressing information is passed directly in the BMR and is not
   derived from the EA bit field in the End-user IPv6 prefix.

   The MAP IPv6 address is created by concatenating the End-user IPv6
   prefix with the MAP subnet identifier (if the End-user IPv6 prefix is
   shorter than 64 bits) and the interface identifier as specified in
   Section 6.

   The MAP subnet identifier is defined to be the first subnet (s bits
   set to zero).


      r = length of the IPv4 prefix given by the BMR;

      o = length of the EA bit field as given by the BMR;

      p = length of the IPv4 suffix contained in the EA bit field.

   The length r MAY be zero, in which case the complete IPv4 address or
   prefix is encoded in the EA bits.  If only a part of the IPv4
   address / prefix is encoded in the EA bits, the Rule IPv4 prefix is
   provisioned to the CE by other means (e.g., a DHCPv6 option).  To
   create a complete IPv4 address (or prefix), the IPv4 address suffix
   (p) from the EA bits is concatenated with the Rule IPv4 prefix
   (r bits).

   The offset of the EA bit field in the IPv6 address is equal to the
   BMR Rule IPv6 prefix length.  The length of the EA bit field (o) is
   given by the BMR Rule EA-bit length and can be between 0 and 48.  A
   length of 48 means that the complete IPv4 address and port are

Troan, et al.                Standards Track                   [Page 12]

RFC 7597                          MAP-E                        July 2015

   embedded in the End-user IPv6 prefix (a single port is assigned).  A
   length of 0 means that no part of the IPv4 address or port is
   embedded in the address.  The sum of the Rule IPv6 Prefix length and
   the Rule EA-bit length MUST be less than or equal to the End-user
   IPv6 prefix length.

   If o + r < 32 (length of the IPv4 address in bits), then an IPv4
   prefix is assigned.  This case is shown in Figure 4.

                   |   r bits    |  o bits =  p bits   |
                   |  Rule IPv4  | IPv4 address suffix |
                   |           < 32 bits               |

                           Figure 4: IPv4 Prefix

   If o + r is equal to 32, then a full IPv4 address is to be assigned.
   The address is created by concatenating the Rule IPv4 prefix and the
   EA-bits.  This case is shown in Figure 5.

                   |   r bits    |  o bits = p bits    |
                   |  Rule IPv4  | IPv4 address suffix |
                   |            32 bits                |

                      Figure 5: Complete IPv4 Address

   If o + r is > 32, then a shared IPv4 address is to be assigned.  The
   number of IPv4 address suffix bits (p) in the EA bits is given by
   32 - r bits.  The PSID bits are used to create a port set.  The
   length of the PSID bit field within the EA bits is q = o - p.

       |   r bits    |        p bits       |         |   q bits   |
       +-------------+---------------------+         +------------+
       |  Rule IPv4  | IPv4 address suffix |         |Port Set ID |
       +-------------+---------------------+         +------------+
       |            32 bits                |

                       Figure 6: Shared IPv4 Address

   The length of r MAY be 32, with no part of the IPv4 address embedded
   in the EA bits.  This results in a mapping with no dependence between
   the IPv4 address and the IPv6 address.  In addition, the length of o
   MAY be zero (no EA bits embedded in the End-user IPv6 prefix),
   meaning that the PSID is also provisioned using, for example, DHCP.

Troan, et al.                Standards Track                   [Page 13]

RFC 7597                          MAP-E                        July 2015

   See Appendix A for an example of the Basic Mapping Rule.

5.3.  Forwarding Mapping Rule (FMR)

   The Forwarding Mapping Rule is optional and is used in Mesh mode to
   enable direct CE-to-CE connectivity.

   On adding an FMR rule, an IPv4 route is installed in the rule table
   for the Rule IPv4 prefix (Figures 4, 5, and 6).

   |        32 bits           |         |    16 bits        |
   +--------------------------+         +-------------------+
   | IPv4 destination address |         |  IPv4 dest port   |
   +--------------------------+         +-------------------+
                  :           :           ___/       :
                  |  p bits   |          /  q bits   :
                  +-----------+         +------------+
                  |IPv4 suffix|         |Port Set ID |
                  +-----------+         +------------+
                   \          /    ____/    ________/
                     \       :  __/   _____/
                       \     : /     /
   |     n bits         |  o bits   | s bits  |   128-n-o-s bits      |
   |  Rule IPv6 prefix  |  EA bits  |subnet ID|     interface ID      |
   |<---  End-user IPv6 prefix  --->|

                 Figure 7: Derivation of MAP IPv6 Address

   See Appendix A for an example of the Forwarding Mapping Rule.

5.4.  Destinations outside the MAP Domain

   IPv4 traffic between MAP nodes that are all within one MAP domain is
   encapsulated in IPv6, with the sender's MAP IPv6 address as the IPv6
   source address and the receiving MAP node's MAP IPv6 address as the
   IPv6 destination address.  To reach IPv4 destinations outside of the
   MAP domain, traffic is also encapsulated in IPv6, but the destination
   IPv6 address is set to the configured IPv6 address of the MAP BR.

   On the CE, the path to the BR can be represented as a point-to-point
   IPv4-over-IPv6 tunnel [RFC2473] with the source address of the tunnel
   being the CE's MAP IPv6 address and the BR IPv6 address as the remote
   tunnel address.  When MAP is enabled, a typical CE router will
   install a default IPv4 route to the BR.

Troan, et al.                Standards Track                   [Page 14]

RFC 7597                          MAP-E                        July 2015

   The BR forwards traffic received from the outside to CEs using the
   normal MAP forwarding rules.

6.  The IPv6 Interface Identifier

   The interface identifier format of a MAP node is described below.

                   |          128-n-o-s bits          |
                   | 16 bits|    32 bits     | 16 bits|
                   |   0    |  IPv4 address  |  PSID  |

                    Figure 8: IPv6 Interface Identifier

   In the case of an IPv4 prefix, the IPv4 address field is right-padded
   with zeros up to 32 bits.  The PSID field is left-padded with zeros
   to create a 16-bit field.  For an IPv4 prefix or a complete IPv4
   address, the PSID field is zero.

   If the End-user IPv6 prefix length is larger than 64, the most
   significant parts of the interface identifier are overwritten by the

7.  MAP Configuration

   For a given MAP domain, the BR and CE MUST be configured with the
   following MAP elements.  The configured values for these elements are
   identical for all CEs and BRs within a given MAP domain.

   o  The Basic Mapping Rule and, optionally, the Forwarding Mapping
      Rules, including the Rule IPv6 prefix, Rule IPv4 prefix, and
      Length of EA bits.

   o  Hub-and-spoke mode or Mesh mode (if all traffic should be sent to
      the BR, or if direct CE-to-CE traffic should be supported).

   In addition, the MAP CE MUST be configured with the IPv6 address(es)
   of the MAP BR (Section 5.4).

7.1.  MAP CE

   The MAP elements are set to values that are the same across all CEs
   within a MAP domain.  The values may be configured in a variety of
   ways, including provisioning methods such as the Broadband Forum's
   "TR-69" Residential Gateway management interface [TR069], an
   XML-based object retrieved after IPv6 connectivity is established, or
   manual configuration by an administrator.  IPv6 DHCP options for MAP

Troan, et al.                Standards Track                   [Page 15]

RFC 7597                          MAP-E                        July 2015

   configuration are defined in [RFC7598].  Other configuration and
   management methods may use the formats described by these options for
   consistency and convenience of implementation on CEs that support
   multiple configuration methods.

   The only remaining provisioning information the CE requires in order
   to calculate the MAP IPv4 address and enable IPv4 connectivity is the
   IPv6 prefix for the CE.  The End-user IPv6 prefix is configured as
   part of obtaining IPv6 Internet access.

   The MAP provisioning parameters, and hence the IPv4 service itself,
   are tied to the associated End-user IPv6 prefix lifetime; thus, the
   MAP service is also tied to this in terms of authorization,
   accounting, etc.

   A single MAP CE MAY be connected to more than one MAP domain, just as
   any router may have more than one IPv4-enabled service-provider-
   facing interface and more than one set of associated addresses
   assigned by DHCP.  Each domain within which a given CE operates would
   require its own set of MAP configuration elements and would generate
   its own IPv4 address.  Each MAP domain requires a distinct End-user
   IPv6 prefix.

   MAP DHCP options are specified in [RFC7598].

7.2.  MAP BR

   The MAP BR MUST be configured with corresponding mapping rules for
   each MAP domain for which it is acting as a BR.

   For increased reliability and load balancing, the BR IPv6 address MAY
   be an anycast address shared across a given MAP domain.  As MAP is
   stateless, any BR may be used at any time.  If the BR IPv6 address is
   anycast, the relay MUST use this anycast IPv6 address as the source
   address in packets relayed to CEs.

   Since MAP uses provider address space, no specific routes need to be
   advertised externally for MAP to operate in IPv6 or IPv4 BGP.
   However, if anycast is used for the MAP IPv6 relays, the anycast
   addresses must be advertised in the service provider's IGP.

Troan, et al.                Standards Track                   [Page 16]

RFC 7597                          MAP-E                        July 2015

8.  Forwarding Considerations

   Figure 1 depicts the overall MAP architecture with IPv4 users
   connected to a routed IPv6 network.

   MAP uses encapsulation mode as specified in [RFC2473].

   For a shared IPv4 address, a MAP CE forwarding IPv4 packets from the
   LAN performs NAT44 functions first and creates appropriate NAT44
   bindings.  The resulting IPv4 packets MUST contain the source IPv4
   address and source transport identifiers specified by the MAP
   provisioning parameters.  The IPv4 packet is forwarded using the CE's
   MAP forwarding function.  The IPv6 source and destination addresses
   MUST then be derived as per Section 5 of this document.

8.1.  Receiving Rules

   A MAP CE receiving an IPv6 packet to its MAP IPv6 address sends this
   packet to the CE's MAP function, where it is decapsulated.  The
   resulting IPv4 packet is then forwarded to the CE's NAT44 function,
   where it is handled according to the NAT's translation table.

   A MAP BR receiving IPv6 packets selects a best matching MAP domain
   rule (Rule IPv6 prefix) based on a longest address match of the
   packet's IPv6 source address, as well as a match of the packet
   destination address against the configured BR IPv6 address(es).  The
   selected MAP Rule allows the BR to determine the EA-bits from the
   source IPv6 address.

   To prevent spoofing of IPv4 addresses, any MAP node (CE and BR) MUST
   perform the following validation upon reception of a packet.  First,
   the embedded IPv4 address or prefix, as well as the PSID (if any),
   are extracted from the source IPv6 address using the matching MAP
   Rule.  These represent the range of what is acceptable as source IPv4
   address and port.  Second, the node extracts the source IPv4 address
   and port from the IPv4 packet encapsulated inside the IPv6 packet.
   If they are found to be outside the acceptable range, the packet MUST
   be silently discarded and a counter incremented to indicate that a
   potential spoofing attack may be underway.  The source validation
   checks just described are not done for packets whose source IPv6
   address is that of the BR (BR IPv6 address).

   By default, the CE router MUST drop packets received on the MAP
   virtual interface (i.e., after decapsulation of IPv6) for IPv4
   destinations not for its own IPv4 shared address, full IPv4 address,
   or IPv4 prefix.

Troan, et al.                Standards Track                   [Page 17]

RFC 7597                          MAP-E                        July 2015

8.2.  ICMP

   ICMP messages should be supported in MAP domains.  Hence, the NAT44
   in the MAP CE MUST implement the behavior for ICMP messages
   conforming to the best current practice documented in [RFC5508].

   If a MAP CE receives an ICMP message having the ICMP Identifier field
   in the ICMP header, the NAT44 in the MAP CE MUST rewrite this field
   to a specific value assigned from the port set.  BRs and other CEs
   must handle this field in a way similar to the handling of a port
   number in the TCP/UDP header upon receiving the ICMP message with the
   ICMP Identifier field.

   If a MAP node receives an ICMP error message without the ICMP
   Identifier field for errors that are detected inside an IPv6 tunnel,
   a node should relay the ICMP error message to the original source.
   This behavior SHOULD be implemented in accordance with Section 8 of

8.3.  Fragmentation and Path MTU Discovery

   Due to the different sizes of the IPv4 and IPv6 headers, handling the
   maximum packet size is relevant for the operation of any system
   connecting the two address families.  There are three mechanisms to
   handle this issue: Path MTU Discovery (PMTUD), fragmentation, and
   transport-layer negotiation such as the TCP Maximum Segment Size
   (MSS) option [RFC879].  MAP uses all three mechanisms to deal with
   different cases.

8.3.1.  Fragmentation in the MAP Domain

   Encapsulating an IPv4 packet to carry it across the MAP domain will
   increase its size (typically by 40 bytes).  It is strongly
   recommended that the MTU in the MAP domain be well managed and that
   the IPv6 MTU on the CE WAN-side interface be set so that no
   fragmentation occurs within the boundary of the MAP domain.

   For an IPv4 packet entering a MAP domain, fragmentation is performed
   as described in Section 7.2 of [RFC2473].

   The use of an anycast source address could lead to an ICMP error
   message generated on the path being sent to a different BR.
   Therefore, using a dynamically set tunnel MTU (Section 6.7 of
   [RFC2473]) is subject to IPv6 Path MTU black holes.  A MAP BR using
   an anycast source address SHOULD NOT by default use Path MTU
   Discovery across the MAP domain.

Troan, et al.                Standards Track                   [Page 18]

RFC 7597                          MAP-E                        July 2015

   Multiple BRs using the same anycast source address could send
   fragmented packets to the same CE at the same time.  If the
   fragmented packets from different BRs happen to use the same
   fragment ID, incorrect reassembly might occur.  See [RFC4459] for an
   analysis of the problem; Section 3.4 of [RFC4459] suggests solving
   the problem by fragmenting the inner packet.

8.3.2.  Receiving IPv4 Fragments on the MAP Domain Borders

   The forwarding of an IPv4 packet received from outside of the MAP
   domain requires the IPv4 destination address and the
   transport-protocol destination port.  The transport-protocol
   information is only available in the first fragment received.  As
   described in Section 5.3.3 of [RFC6346], a MAP node receiving an
   IPv4 fragmented packet from outside has to reassemble the packet
   before sending the packet onto the MAP link.  If the first packet
   received contains the transport-protocol information, it is possible
   to optimize this behavior by using a cache and forwarding the
   fragments unchanged.  Implementers of MAP should be aware that there
   are a number of well-known attacks against IP fragmentation; see
   [RFC1858] and [RFC3128].  Implementers should also be aware of
   additional issues with reassembling packets at high rates, as
   described in [RFC4963].

8.3.3.  Sending IPv4 Fragments to the Outside

   If two IPv4 hosts behind two different MAP CEs with the same IPv4
   address send fragments to an IPv4 destination host outside the
   domain, those hosts may use the same IPv4 fragmentation identifier,
   resulting in incorrect reassembly of the fragments at the destination
   host.  Given that the IPv4 fragmentation identifier is a 16-bit
   field, it could be used similarly to port ranges.  A MAP CE could
   rewrite the IPv4 fragmentation identifier to be within its allocated
   port set, if the resulting fragment identifier space was large enough
   related to the rate at which fragments were sent.  However, splitting
   the identifier space in this fashion would increase the probability
   of reassembly collisions for all connections through the Customer
   Premises Equipment (CPE).  See also [RFC6864].

9.  NAT44 Considerations

   The NAT44 implemented in the MAP CE SHOULD conform to the behavior
   and best current practices documented in [RFC4787], [RFC5508], and
   [RFC5382].  In MAP address-sharing mode (determined by the MAP
   domain / rule configuration parameters), the operation of the NAT44
   MUST be restricted to the available port numbers derived via the
   Basic Mapping Rule.

Troan, et al.                Standards Track                   [Page 19]

RFC 7597                          MAP-E                        July 2015

10.  Security Considerations

   Spoofing attacks:  With consistency checks between IPv4 and IPv6
      sources that are performed on IPv4/IPv6 packets received by MAP
      nodes, MAP does not introduce any new opportunity for spoofing
      attacks that would not already exist in IPv6.

   Denial-of-service attacks:  In MAP domains where IPv4 addresses are
      shared, the fact that IPv4 datagram reassembly may be necessary
      introduces an opportunity for DoS attacks.  This is inherent in
      address sharing and is common with other address-sharing
      approaches such as DS-Lite and NAT64/DNS64.  The best protection
      against such attacks is to accelerate IPv6 deployment so that
      address sharing is used less and less where MAP is supported.

   Routing loop attacks:  Routing loop attacks may exist in some
      "automatic tunneling" scenarios and are documented in [RFC6324].
      They cannot exist with MAP because each BR checks that the IPv6
      source address of a received IPv6 packet is a CE address based on
      the Forwarding Mapping Rule.

   Attacks facilitated by restricted port set:  From hosts that are not
      subject to ingress filtering [RFC2827], an attacker can inject
      spoofed packets during ongoing transport connections [RFC4953]
      [RFC5961] [RFC6056].  The attacks depend on guessing which ports
      are currently used by target hosts.  Using an unrestricted port
      set is preferable, i.e., using native IPv6 connections that are
      not subject to MAP port-range restrictions.  To minimize these
      types of attacks when using a restricted port set, the MAP CE's
      NAT44 filtering behavior SHOULD be "Address-Dependent Filtering"
      as described in Section 5 of [RFC4787].  Furthermore, the MAP CEs
      SHOULD use a DNS transport proxy [RFC5625] function to handle DNS
      traffic and source such traffic from IPv6 interfaces not assigned
      to MAP.

   [RFC6269] outlines general issues with IPv4 address sharing.

Troan, et al.                Standards Track                   [Page 20]

RFC 7597                          MAP-E                        July 2015

11.  References

11.1.  Normative References

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

   [RFC2473]  Conta, A. and S. Deering, "Generic Packet Tunneling in
              IPv6 Specification", RFC 2473, DOI 10.17487/RFC2473,
              December 1998, <http://www.rfc-editor.org/info/rfc2473>.

   [RFC5625]  Bellis, R., "DNS Proxy Implementation Guidelines",
              BCP 152, RFC 5625, DOI 10.17487/RFC5625, August 2009,

11.2.  Informative References

              Sun, Q., Chen, M., Chen, G., Tsou, T., and S. Perreault,
              "Mapping of Address and Port (MAP) - Deployment
              Considerations", Work in Progress,
              draft-ietf-softwire-map-deployment-06, June 2015.

   [RFC879]   Postel, J., "The TCP Maximum Segment Size and Related
              Topics", RFC 879, DOI 10.17487/RFC0879, November 1983,

   [RFC1858]  Ziemba, G., Reed, D., and P. Traina, "Security
              Considerations for IP Fragment Filtering", RFC 1858,
              DOI 10.17487/RFC1858, October 1995,

   [RFC1933]  Gilligan, R. and E. Nordmark, "Transition Mechanisms for
              IPv6 Hosts and Routers", RFC 1933, DOI 10.17487/RFC1933,
              April 1996, <http://www.rfc-editor.org/info/rfc1933>.

   [RFC2529]  Carpenter, B. and C. Jung, "Transmission of IPv6 over IPv4
              Domains without Explicit Tunnels", RFC 2529,
              DOI 10.17487/RFC2529, March 1999,

   [RFC2663]  Srisuresh, P. and M. Holdrege, "IP Network Address
              Translator (NAT) Terminology and Considerations",
              RFC 2663, DOI 10.17487/RFC2663, August 1999,

Troan, et al.                Standards Track                   [Page 21]

RFC 7597                          MAP-E                        July 2015

   [RFC2827]  Ferguson, P. and D. Senie, "Network Ingress Filtering:
              Defeating Denial of Service Attacks which employ IP Source
              Address Spoofing", BCP 38, RFC 2827, DOI 10.17487/RFC2827,
              May 2000, <http://www.rfc-editor.org/info/rfc2827>.

   [RFC3056]  Carpenter, B. and K. Moore, "Connection of IPv6 Domains
              via IPv4 Clouds", RFC 3056, DOI 10.17487/RFC3056,
              February 2001, <http://www.rfc-editor.org/info/rfc3056>.

   [RFC3128]  Miller, I., "Protection Against a Variant of the Tiny
              Fragment Attack (RFC 1858)", RFC 3128,
              DOI 10.17487/RFC3128, June 2001,

   [RFC3633]  Troan, O. and R. Droms, "IPv6 Prefix Options for Dynamic
              Host Configuration Protocol (DHCP) version 6", RFC 3633,
              DOI 10.17487/RFC3633, December 2003,

   [RFC4213]  Nordmark, E. and R. Gilligan, "Basic Transition Mechanisms
              for IPv6 Hosts and Routers", RFC 4213,
              DOI 10.17487/RFC4213, October 2005,

   [RFC4459]  Savola, P., "MTU and Fragmentation Issues with
              In-the-Network Tunneling", RFC 4459, DOI 10.17487/RFC4459,
              April 2006, <http://www.rfc-editor.org/info/rfc4459>.

   [RFC4632]  Fuller, V. and T. Li, "Classless Inter-domain Routing
              (CIDR): The Internet Address Assignment and Aggregation
              Plan", BCP 122, RFC 4632, DOI 10.17487/RFC4632,
              August 2006, <http://www.rfc-editor.org/info/rfc4632>.

   [RFC4787]  Audet, F., Ed., and C. Jennings, "Network Address
              Translation (NAT) Behavioral Requirements for Unicast
              UDP", BCP 127, RFC 4787, DOI 10.17487/RFC4787,
              January 2007, <http://www.rfc-editor.org/info/rfc4787>.

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

   [RFC4953]  Touch, J., "Defending TCP Against Spoofing Attacks",
              RFC 4953, DOI 10.17487/RFC4953, July 2007,

Troan, et al.                Standards Track                   [Page 22]

RFC 7597                          MAP-E                        July 2015

   [RFC4963]  Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly
              Errors at High Data Rates", RFC 4963,
              DOI 10.17487/RFC4963, July 2007,

   [RFC5214]  Templin, F., Gleeson, T., and D. Thaler, "Intra-Site
              Automatic Tunnel Addressing Protocol (ISATAP)", RFC 5214,
              DOI 10.17487/RFC5214, March 2008,

   [RFC5382]  Guha, S., Ed., Biswas, K., Ford, B., Sivakumar, S., and P.
              Srisuresh, "NAT Behavioral Requirements for TCP", BCP 142,
              RFC 5382, DOI 10.17487/RFC5382, October 2008,

   [RFC5508]  Srisuresh, P., Ford, B., Sivakumar, S., and S. Guha, "NAT
              Behavioral Requirements for ICMP", BCP 148, RFC 5508,
              DOI 10.17487/RFC5508, April 2009,

   [RFC5961]  Ramaiah, A., Stewart, R., and M. Dalal, "Improving TCP's
              Robustness to Blind In-Window Attacks", RFC 5961,
              DOI 10.17487/RFC5961, August 2010,

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

   [RFC6056]  Larsen, M. and F. Gont, "Recommendations for
              Transport-Protocol Port Randomization", BCP 156, RFC 6056,
              DOI 10.17487/RFC6056, January 2011,

   [RFC6250]  Thaler, D., "Evolution of the IP Model", RFC 6250,
              DOI 10.17487/RFC6250, May 2011,

   [RFC6269]  Ford, M., Ed., Boucadair, M., Durand, A., Levis, P., and
              P. Roberts, "Issues with IP Address Sharing", RFC 6269,
              DOI 10.17487/RFC6269, June 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,

Troan, et al.                Standards Track                   [Page 23]

RFC 7597                          MAP-E                        July 2015

   [RFC6333]  Durand, A., Droms, R., Woodyatt, J., and Y. Lee,
              "Dual-Stack Lite Broadband Deployments Following IPv4
              Exhaustion", RFC 6333, DOI 10.17487/RFC6333, August 2011,

   [RFC6335]  Cotton, M., Eggert, L., Touch, J., Westerlund, M., and S.
              Cheshire, "Internet Assigned Numbers Authority (IANA)
              Procedures for the Management of the Service Name and
              Transport Protocol Port Number Registry", BCP 165,
              RFC 6335, DOI 10.17487/RFC6335, 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,

   [RFC6864]  Touch, J., "Updated Specification of the IPv4 ID Field",
              RFC 6864, DOI 10.17487/RFC6864, February 2013,

   [RFC7598]  Mrugalski, T., Troan, O., Farrer, I., Perreault, S., Dec,
              W., Bao, C., Yeh, L., and X. Deng, "DHCPv6 Options for
              Configuration of Softwire Address and Port-Mapped
              Clients", RFC 7598, DOI 10.17487/RFC7598, July 2015,

              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.

   [TR069]    Broadband Forum TR-069, "CPE WAN Management Protocol",
              Amendment 5, CWMP Version: 1.4, November 2013,

Troan, et al.                Standards Track                   [Page 24]

RFC 7597                          MAP-E                        July 2015

Appendix A.  Examples

   Example 1 - Basic Mapping Rule:

   Given the MAP domain information and an IPv6 address of
   an endpoint:

   End-user IPv6 prefix: 2001:db8:0012:3400::/56
   Basic Mapping Rule:   {2001:db8:0000::/40 (Rule IPv6 prefix),
                 (Rule IPv4 prefix),
                          16 (Rule EA-bit length)}
   PSID length:          (16 - (32 - 24) = 8 (sharing ratio of 256)
   PSID offset:          6 (default)

   A MAP node (CE or BR) can, via the BMR or equivalent FMR,
   determine the IPv4 address and port set as shown below:

   EA bits offset:       40
   IPv4 suffix bits (p)  Length of IPv4 address (32) -
                         IPv4 prefix length (24) = 8
   IPv4 address: (0xc0000212)
   PSID start:           40 + p = 40 + 8 = 48
   PSID length:          o - p = (56 - 40) - 8 = 8
   PSID:                 0x34

   Available ports (63 ranges): 1232-1235, 2256-2259, ...... ,
                                63696-63699, 64720-64723

   The BMR information allows a MAP CE to determine (complete)
   its IPv6 address within the indicated IPv6 prefix.

   IPv6 address of MAP CE:  2001:db8:0012:3400:0000:c000:0212:0034

Troan, et al.                Standards Track                   [Page 25]

RFC 7597                          MAP-E                        July 2015

   Example 2 - BR:

   Another example is a MAP BR, configured with the following FMR
   when receiving a packet with the following characteristics:

   IPv4 source address: (0x01020304)
   IPv4 source port:          80
   IPv4 destination address: (0xc0000212)
   IPv4 destination port:     1232

   Forwarding Mapping Rule: {2001:db8::/40 (Rule IPv6 prefix),
                    (Rule IPv4 prefix),
                             16 (Rule EA-bit length)}

   IPv6 address of MAP BR:              2001:db8:ffff::1

   The above information allows the BR to derive the mapped
   destination IPv6 address for the corresponding MAP CE, and also
   the mapped source IPv6 address for the IPv4 source address,
   as follows:

   IPv4 suffix bits (p):  32 - 24 = 8 (18 (0x12))
   PSID length:           8
   PSID:                  0x34 (1232)

   The resulting IPv6 packet will have the following key fields:

   IPv6 source address:       2001:db8:ffff::1
   IPv6 destination address:  2001:db8:0012:3400:0000:c000:0212:0034

   Example 3 - Forwarding Mapping Rule:

   An IPv4 host behind the MAP CE (addressed as per the previous
   examples) corresponding with IPv4 host will have its
   packets encapsulated by IPv6 using the IPv6 address of the BR
   configured on the MAP CE as follows:

   IPv6 address of BR:         2001:db8:ffff::1
   IPv4 source address:
   IPv4 destination address:
   IPv4 source port:           1232
   IPv4 destination port:      80
   MAP CE IPv6 source address: 2001:db8:0012:3400:0000:c000:0212:0034
   IPv6 destination address:   2001:db8:ffff::1

Troan, et al.                Standards Track                   [Page 26]

RFC 7597                          MAP-E                        July 2015

   Example 4 - Rule with no embedded address bits and no address

   End-user IPv6 prefix: 2001:db8:0012:3400::/56
   Basic Mapping Rule:   {2001:db8:0012:3400::/56 (Rule IPv6 prefix),
                 (Rule IPv4 prefix),
                          0 (Rule EA-bit length)}
   PSID length:          0 (sharing ratio is 1)
   PSID offset:          n/a

   A MAP node (CE or BR) can, via the BMR or equivalent FMR, determine
   the IPv4 address and port set as shown below:

   EA bits offset:       0
   IPv4 suffix bits (p): Length of IPv4 address (32) -
                         IPv4 prefix length (32) = 0
   IPv4 address: (0xc0000212)
   PSID start:           0
   PSID length:          0
   PSID:                 null

   The BMR information allows a MAP CE to also determine (complete)
   its full IPv6 address by combining the IPv6 prefix with the MAP
   interface identifier (that embeds the IPv4 address).

   IPv6 address of MAP CE:  2001:db8:0012:3400:0000:c000:0212:0000

Troan, et al.                Standards Track                   [Page 27]

RFC 7597                          MAP-E                        July 2015

   Example 5 - Rule with no embedded address bits and address sharing
   (sharing ratio of 256):

   End-user IPv6 prefix: 2001:db8:0012:3400::/56
   Basic Mapping Rule:   {2001:db8:0012:3400::/56 (Rule IPv6 prefix),
                 (Rule IPv4 prefix),
                          0 (Rule EA-bit length)}
   PSID length:          8 (from DHCP; sharing ratio of 256)
   PSID offset:          6 (default)
   PSID:                 0x34 (from DHCP)

   A MAP node can, via the Basic Mapping Rule, determine the IPv4
   address and port set as shown below:

   EA bits offset:        0
   IPv4 suffix bits (p):  Length of IPv4 address (32) -
                          IPv4 prefix length (32) = 0
   IPv4 address: (0xc0000212)
   PSID offset:           6
   PSID length:           8
   PSID:                  0x34

   Available ports (63 ranges): 1232-1235, 2256-2259, ...... ,
                                63696-63699, 64720-64723

   The Basic Mapping Rule information allows a MAP CE to also
   determine (complete) its full IPv6 address by combining the IPv6
   prefix with the MAP interface identifier (that embeds the IPv4
   address and PSID).

   IPv6 address of MAP CE: 2001:db8:0012:3400:0000:c000:0212:0034

   Note that the IPv4 address and PSID are not derived from the IPv6
   prefix assigned to the CE but are provisioned separately using,
   for example, DHCP.

Troan, et al.                Standards Track                   [Page 28]

RFC 7597                          MAP-E                        July 2015

Appendix B.  A More Detailed Description of the Derivation of the
             Port-Mapping Algorithm

   This appendix describes how the port-mapping algorithm described in
   Section 5.1 was derived.  The algorithm is used in domains whose
   rules allow IPv4 address sharing.

   The basic requirement for a port-mapping algorithm is that the port
   sets it assigns to different MAP CEs MUST be non-overlapping.  A
   number of other requirements guided the choice of the algorithm:

   o  In keeping with the general MAP algorithm, the port set MUST be
      derivable from a Port Set identifier (PSID) that can be embedded
      in the End-user IPv6 prefix.

   o  The mapping MUST be reversible such that, given the port number,
      the PSID of the port set to which it belongs can be quickly

   o  The algorithm MUST allow a broad range of address-sharing ratios.

   o  It SHOULD be possible to exclude subsets of the complete port
      numbering space from assignment.  Most operators would exclude the
      system ports (0-1023).  A conservative operator might exclude all
      but the transient ports (49152-65535).

   o  The effect of port exclusion on the possible values of the
      End-user IPv6 prefix (i.e., due to restrictions on the PSID value)
      SHOULD be minimized.

   o  For administrative simplicity, the algorithm SHOULD allocate the
      same or almost the same number of ports to each CE sharing a given
      IPv4 address.

   The two extreme cases that an algorithm satisfying those conditions
   might support are when (1) the port numbers are not contiguous for
   each PSID but uniformly distributed across the allowed port range and
   (2) the port numbers are contiguous in a single range for each PSID.
   The port-mapping algorithm proposed here is called the Generalized
   Modulus Algorithm (GMA) and supports both of these cases.

Troan, et al.                Standards Track                   [Page 29]

RFC 7597                          MAP-E                        July 2015

   For a given IPv4 address-sharing ratio (R) and the maximum number of
   contiguous ports (M) in a port set, the GMA is defined as follows:

   a.  The port numbers (P) corresponding to a given PSID are
       generated by:

       (1) ... P = (R * M) * i + M * PSID + j

       where i and j are indices and the ranges of i, j, and the PSID
       are discussed below.

   b.  For any given port number P, the PSID is calculated as:

       (2) ... PSID = trunc((P modulo (R * M)) / M)

       where trunc() is the operation of rounding down to the nearest

   Formula (1) can be interpreted as follows.  First, the available port
   space is divided into blocks of size R * M.  Each block is divided
   into R individual ranges of length M.  The index i in formula (1)
   selects a block, PSID selects a range within that block, and the
   index j selects a specific port value within the range.  On the basis
   of this interpretation:

   o  i ranges from ceil(N / (R * M)) to trunc(65536/(R * M)) - 1, where
      ceil is the operation of rounding up to the nearest integer and N
      is the number of ports (e.g., 1024) excluded from the lower end of
      the range.  That is, any block containing excluded values is
      discarded at the lower end, and if the final block has fewer than
      R * M values it is discarded.  This ensures that the same number
      of ports is assigned to every PSID.

   o  PSID ranges from 0 to R - 1.

   o  j ranges from 0 to M - 1.

Troan, et al.                Standards Track                   [Page 30]

RFC 7597                          MAP-E                        July 2015

B.1.  Bit Representation of the Algorithm

   If R and M are powers of 2 (R = 2^k, M = 2^m), formula (1) translates
   to a computationally convenient structure for any port number
   represented as a 16-bit binary number.  This structure is shown in
   Figure 9.

          0                          8                         15
          |                     P                               |
          |        i      |       PSID      |        j          |
          |<----a bits--->|<-----k bits---->|<------m bits----->|

               Figure 9: Bit Representation of a Port Number

   As shown in the figure, the index value i of formula (1) is given by
   the first a = 16 - k - m bits of the port number.  The PSID value is
   given by the next k bits, and the index value j is given by the last
   m bits.

   Because the PSID is always in the same position in the port number
   and always the same length, different PSID values are guaranteed to
   generate different sets of port numbers.  In the reverse direction,
   the generating PSID can be extracted from any port number by a
   bitmask operation.

   Note that when M and R are powers of 2, 65536 divides evenly by
   R * M.  Hence, the final block is complete, and the upper bound on i
   is exactly 65536/(R * M) - 1.  The lower bound on i is still the
   minimum required to ensure that the required set of ports is
   excluded.  No port numbers are wasted through the discarding of
   blocks at the lower end if block size R * M is a factor of N, the
   number of ports to be excluded.

   As a final note, the number of blocks into which the range 0-65535 is
   being divided in the above representation is given by 2^a.  Hence,
   the case where a = 0 can be interpreted as one where the complete
   range has been divided into a single block, and individual port sets
   are contained in contiguous ranges in that block.  We cannot throw
   away the whole block in that case, so port exclusion has to be
   achieved by putting a lower bound equal to ceil(N / M) on the allowed
   set of PSID values instead.

Troan, et al.                Standards Track                   [Page 31]

RFC 7597                          MAP-E                        July 2015

B.2.  GMA Examples

   For example, for R = 256, PSID = 0, offset: a = 6 and PSID length:
   k = 8 bits:

   Available ports (63 ranges): 1024-1027, 2048-2051, ...... ,
                                63488-63491, 64512-64515

                    Example 1: with offset = 6 (a = 6)

   For example, for R = 64, PSID = 0, a = 0 (PSID offset = 0 and PSID
   length = 6 bits), no port exclusion:

   Available ports (1 range): 0-1023

               Example 2: with offset = 0 (a = 0) and N = 0


   This document is based on the ideas of many, including Masakazu
   Asama, Mohamed Boucadair, Gang Chen, Maoke Chen, Wojciech Dec,
   Xiaohong Deng, Jouni Korhonen, Tomek Mrugalski, Jacni Qin, Chunfa
   Sun, Qiong Sun, and Leaf Yeh.  The authors want in particular to
   recognize Remi Despres, who has tirelessly worked on generalized
   mechanisms for stateless address mapping.

   The authors would like to thank Lichun Bao, Guillaume Gottard, Dan
   Wing, Jan Zorz, Necj Scoberne, Tina Tsou, Kristian Poscic, and
   especially Tom Taylor and Simon Perreault for the thorough review and
   comments of this document.  Useful IETF Last Call comments were
   received from Brian Weis and Lei Yan.

Troan, et al.                Standards Track                   [Page 32]

RFC 7597                          MAP-E                        July 2015


   This document is the result of the IETF Softwire MAP design team
   effort and numerous previous individual contributions in this area:

   Chongfeng Xie
   China Telecom
   Room 708, No. 118, Xizhimennei Street
   Beijing  100035
   Phone: +86-10-58552116
   Email: xiechf@ctbri.com.cn

   Qiong Sun
   China Telecom
   Room 708, No. 118, Xizhimennei Street
   Beijing  100035
   Phone: +86-10-58552936
   Email: sunqiong@ctbri.com.cn

   Gang Chen
   China Mobile
   29, Jinrong Avenue
   Xicheng District, Beijing  100033
   Email: phdgang@gmail.com, chengang@chinamobile.com

   Yu Zhai
   CERNET Center/Tsinghua University
   Room 225, Main Building, Tsinghua University
   Beijing  100084
   Email: jacky.zhai@gmail.com

   Wentao Shang
   CERNET Center/Tsinghua University
   Room 225, Main Building, Tsinghua University
   Beijing  100084
   Email: wentaoshang@gmail.com

Troan, et al.                Standards Track                   [Page 33]

RFC 7597                          MAP-E                        July 2015

   Guoliang Han
   CERNET Center/Tsinghua University
   Room 225, Main Building, Tsinghua University
   Beijing  100084
   Email: bupthgl@gmail.com

   Rajiv Asati
   Cisco Systems
   7025-6 Kit Creek Road
   Research Triangle Park, NC  27709
   United States
   Email: rajiva@cisco.com

Authors' Addresses

   Ole Troan (editor)
   Cisco Systems
   Philip Pedersens vei 1
   Lysaker  1366

   Email: ot@cisco.com

   Wojciech Dec
   Cisco Systems
   Haarlerbergpark Haarlerbergweg 13-19
   Amsterdam, NOORD-HOLLAND  1101 CH
   The Netherlands

   Email: wdec@cisco.com

   Xing Li
   CERNET Center/Tsinghua University
   Room 225, Main Building, Tsinghua University
   Beijing  100084

   Email: xing@cernet.edu.cn

Troan, et al.                Standards Track                   [Page 34]

RFC 7597                          MAP-E                        July 2015

   Congxiao Bao
   CERNET Center/Tsinghua University
   Room 225, Main Building, Tsinghua University
   Beijing  100084

   Email: congxiao@cernet.edu.cn

   Satoru Matsushima
   SoftBank Telecom
   1-9-1 Higashi-Shinbashi, Munato-ku

   Email: satoru.matsushima@g.softbank.co.jp

   Tetsuya Murakami
   IP Infusion
   1188 East Arques Avenue
   Sunnyvale, CA  94085
   United States

   Email: tetsuya@ipinfusion.com

   Tom Taylor (editor)
   Huawei Technologies

   Email: tom.taylor.stds@gmail.com

Troan, et al.                Standards Track                   [Page 35]