<?xml version="1.0" encoding="UTF-8"?>
  <?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
  <!-- generated by https://github.com/cabo/kramdown-rfc version 1.7.35 (Ruby 4.0.2) -->


<!DOCTYPE rfc  [
  <!ENTITY nbsp    "&#160;">
  <!ENTITY zwsp   "&#8203;">
  <!ENTITY nbhy   "&#8209;">
  <!ENTITY wj     "&#8288;">

]>

<?rfc toc, sortrefs, symrefs, comments="yes"?>

<rfc ipr="trust200902" docName="draft-ietf-idr-multinexthop-attribute-06" category="std" consensus="true" submissionType="IETF">
  <front>
    <title abbrev="BGP MNH attribute">BGP MultiNexthop Attribute</title>

    <author initials="K." surname="Vairavakkalai" fullname="Kaliraj Vairavakkalai">
      <organization>HPE</organization>
      <address>
        <postal>
          <street>1133 Innovation Way</street>
          <city>Sunnyvale</city>
          <region>CA</region>
          <code>94089</code>
          <country>USA</country>
        </postal>
        <email>kaliraj@hpe.com</email>
      </address>
    </author>
    <author initials="M." surname="Nanduri" fullname="Mohan Nanduri">
      <organization>Microsoft</organization>
      <address>
        <postal>
          <street>1 Microsoft Way</street>
          <city>Redmond</city>
          <region>WA</region>
          <code>98052</code>
          <country>USA</country>
        </postal>
        <email>mohannanduri@microsoft.com</email>
      </address>
    </author>
    <author initials="A." surname="Reddy" fullname="Avinash Reddy">
      <organization>AT&amp;T</organization>
      <address>
        <postal>
          <street>400 W Plano Pkwy</street>
          <city>Plano</city>
          <region>TX</region>
          <code>75075</code>
          <country>USA</country>
        </postal>
        <email>ar977m@att.com</email>
      </address>
    </author>

    <date year="2026" month="November" day="19"/>

    <area>Routing</area>
    <workgroup>IDR Working Group</workgroup>
    <keyword>BGP MNH, MultiNexthop Attribute</keyword>

    <abstract>


<?line 109?>

<t>Today, a BGP speaker can advertise one nexthop for a set of NLRIs in an Update message. This nexthop can be encoded in either the top-level BGP-Nexthop attribute (code 3), or inside the MP_REACH_NLRI attribute (code 14). Forwarding information related to the nexthop is scattered across various attributes, extended communities or the NLRI field.</t>

<t>This document defines a new optional non-transitive BGP attribute called "MultiNexthop (MNH)" with IANA code TBD. The MNH provides two things: it allows carrying the Nexthop and related forwarding information in one BGP attribute. The MNH also enables carrying an ordered set of multiple Nexthops in the same attribute, with forwarding information scoped on a per Nexthop basis.</t>



    </abstract>



  </front>

  <middle>


<?line 116?>

<section anchor="introduction"><name>Introduction</name>

<t>Today, a BGP speaker can advertise one nexthop for a set of NLRIs in an Update message. This nexthop can be encoded in either the top-level BGP-Nexthop attribute (code 3), or inside the MP_REACH_NLRI attribute (code 14). Forwarding information related to the nexthop is scattered across various attributes, extended communities or the NLRI field.</t>

<t>This document defines a new optional non-transitive BGP attribute called "MultiNexthop (MNH)" with IANA code TBD. The MNH provides two things: it allows carrying the Nexthop and related forwarding information in one BGP attribute. The MNH also enables carrying an ordered set of multiple Nexthops in the same attribute, with forwarding information scoped on a per Nexthop basis.</t>

</section>
<section anchor="terminology"><name>Terminology</name>

<t>iSN: Ingress Service Node</t>

<t>eSN: Egress Service Node</t>

<t>NLRI: Network Layer Reachability Information</t>

<t>AFI: Address Family Identifier</t>

<t>SAFI: Subsequent Address Family Identifier</t>

<t>PE : Provider Edge</t>

<t>RT : Route-Target extended community</t>

<t>RD : Route-Distinguisher</t>

<t>MPLS: Multi Protocol Label Switching</t>

<t>ECMP: Equal Cost Multi Path</t>

<t>WECMP: Weighted Equal Cost Multi Path</t>

<t>FRR: Fast Re Route</t>

<t>PNH : Protocol Next hop address carried in a BGP Update message</t>

<t>MNH: BGP MultiNextHop attribute</t>

<t>NFI: Nexthop Forwarding Information</t>

<t>FI: Forwarding Instruction</t>

<t>FA: Forwarding Argument</t>

<section anchor="definitions"><name>Definitions</name>

<t>MULTI_NEXT_HOP (aka MNH): BGP MultiNexthop attribute. The new attribute defined by this document.</t>

<t>MNH TLV: Top level TLV contained in a MULTI_NEXT_HOP.</t>

<t>NFI TLV: Nexthop Forwarding Information TLV, contained in a MNH TLV.</t>

<t>FI TLV: Forwarding Instruction TLV, contained in a NFI TLV.</t>

<t>FA TLV: Forwarding Argument TLV, contained as an argument to a FI in the FI TLV.</t>

</section>
</section>
<section anchor="motivation"><name>Motivation</name>

<t>Today, in a BGP Update, forwarding information related to the BGP nexthop is scattered across various attributes, extended communities or the NLRI field. On some other address families like Flowspec, nexthop address is carried without using the nexthop attribute, in one or more extended communities of specific type.</t>

<t>It may be desirable to carry the forwarding information for a nexthop scoped in a single attribute, and uniformly for all address families.</t>

<t>For cases where multiple nexthops need to be advertised, BGP Addpath <xref target="RFC7911"/> is used with some address families. Though Addpath allows basic ability to advertise multiple routes, it does not allow the sender to express the desired relationship between the multiple nexthops being advertised e.g., relative ordering, type of load balancing, fast reroute. These are local decisions based on configuration and path selection at the receiving node. Also, Addpath does not consider things like Link-bandwidth community when selecting add-path routes. Some scenarios (explained in Appendix A) may benefit from having a mechanism, where egress node can signal multiple nexthops along with their relationship to ingress nodes.</t>

<t>It would be desirable to have a common way to carry more than one nexthop on a BGP route of any family, and express relationship between them.</t>

<t>This document defines a new optional non-transitive BGP attribute "MultiNexthop (MNH)" that can be used for these purposes. The MNH attribute can be used in any BGP family that wants to carry one or more nexthops, with all forwarding information being carried in one attribute, scoped on a per nexthop basis.</t>

<t>E.g. The MNH can be used to advertise MPLS label along with nexthop for labeled and unlabeled families (e.g. Inet Unicast, Inet6 Unicast, Flowspec) alike. Such that, mechanisms at the transport layer can work uniformly on labeled and unlabled BGP families to realize various use-cases.</t>

</section>
<section anchor="spec"><name>Base Encoding And Protocol Procedures</name>

<t>"MultiNexthop (MNH)" is a new BGP optional non-transitive attribute (code TBD), that can be used to carry an ordered set of one or more Nexthops in the same route, with all forwarding information being carried in one attribute, scoped on a per nexthop basis. This attribute describes forwarding instructions using TLVs as shown below.</t>

<t>This section describes the organization and encoding of the MNH attribute.</t>

<figure title="Overview of MNH Attribute Layout - Eye candy summary" anchor="mnh-eye-candy"><artwork><![CDATA[
          MNH Attribute: {
                PrimaryPath {
                    [Forwarding Instruction 1],
                     ..
                    [Forwarding Instruction n]
                }
                RepairPath {
                    [Forwarding Instruction 1],
                     ..
                    [Forwarding Instruction n]
                }
          }

          Forwarding Instruction: {
              {FwdAction, Forwarding Arguments}
          }

]]></artwork></figure>

<t>A MNH attribute consists of a Header and one or more "MNH TLVs".</t>

<t>A MNH TLV contains a Type and one unit of "Nexthop Forwarding Information" (NFI TLV).</t>

<t>A NFI TLV contains one or more "Forwarding Instructions" (FI TLVs).</t>

<t>A FI TLV contains a "Forwarding Action" code and one more "Forwarding Arguments" (FA TLVs).</t>

<t>The FA TLVs describe the parameters required to complete a "Forwarding Action".</t>

<section anchor="multinexthop-attribute"><name>MultiNexthop Attribute</name>

<figure title="MultiNexthop - BGP Attribute" anchor="mnh-attr"><artwork><![CDATA[
    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Attr. Flags  |Attr. Type Code|          Length               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |Ver| MNH-Flags |       Advertising Router ID  (Four ..         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   .. bytes)   |       MNH TLV                                 ~
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   ~                       MNH TLV                                 |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


 MNH Header:
   - Attr. Flags (1 octet)
       BGP Path-attribute flags. indicating an Optional Non-Transitive
       attribute. i.e. Optional bit set, Transitive bit reset.

   - Attr. Type Code (1 octet)
       Type code allotted by IANA. TBD.

   - Length (1 or 2 octets)
       One or Two bytes field stating length of attribute value in bytes.

   - Version (2 bits)
       MNH Version - indicates layout of the MNH header.
       Set to 0x0 indicating "MNH v0", which is defined in this document.

       If there is any significant changes to the skeletal layout of
       MNH attribute in future, this Version field will be useful.

   - MNH Flags (6 bits)

           2 3 4 5 6 7
          +-+-+-+-+-+-+
          |R R R R R M|
          +-+-+-+-+-+-+

       6 bits following the Version bits are MNH Flags.

           M: "Mandatory".
              Value 1 indicates that this MNH attribute is mandatory.
              If this MNH attr is invalid, the route is Unusable Hidden.

           R: Reserved. MUST be set to zero, SHOULD be ignored by receiver.

   - Advertising Router ID (4 octets)
      Router ID of the BGP speaker that is the source of truth for the
      contents of this MNH attribute.

 MNH TLVs: One or more MNH TLVs are carried in a MNH attr.
      MNH TLV is described in subsequent sections.

]]></artwork></figure>

<section anchor="processing-the-mnh-header"><name>Processing the MNH Header</name>

<t>A BGP speaker MUST fill MNH Version field to 0.</t>

<t>If a MNH is received with a Version other than 0, the MNH attribute MUST be considered invalid, and be treated as Unrecognized Non-transitive attribute.</t>

<t>The "Advertising Router ID" field identifies the router that added the MNH attribute.</t>

<t>The "Advertising Router ID" field is intended to help in debugging and troubleshooting only. It is not used for validating the MNH. The non-transitive nature of MNH avoids attribute escape.</t>

</section>
<section anchor="scope-of-use-origination-and-propagation"><name>Scope of Use, Origination and Propagation</name>

<t>The MNH attribute is intended to be used in a BGP free core, between egress and ingress BGP speakers that understand this attribute. These BGP speakers may have an intra-AS or inter-AS BGP session between them.</t>

<t>To avoid un-intentionally leaking the MNH to another AS, via a BGP speaker that does not understand MNH attribute, it is defined as "optional non-transitive". But this also means that a RR needs to be upgraded to support this attribute before any PEs in the network can make use of it.</t>

<t>Use of MNH on a BGP session is disabled by default. An implementation MUST provide configuration control on a per BGP neighbor address family basis, to enable MNH support.</t>

<t>A BGP speaker MUST NOT advertise MNH on a BGP route if MNH support is not enabled for the corresponding address family on the advertising BGP session.</t>

<t>If the MNH attribute is received on a BGP session where MNH support is not enabled, the attribute MUST be treated as Unrecognized non-transitive. This rule provides additional protection against unintended propagation of this attribute, when both BGP speakers understand MNH but receiver has not enabled the support. A RFC3392 Capability is not used for this purpose, because it would cause session reset whenever 'MNH support' config is changed.</t>

<t>Remaining text in this section apply when both receiving and advertising BGP sessions are enabled with MNH support.</t>

<t>When a BGP speaker receives the MNH attribute on a BGP route, and re-advertises it with the nexthop unchanged, it MUST propagate the attribute unchanged. E.g. a Route Reflector. The "Advertising Router ID" field in the MNH will thus contain the original BGP speaker that added it.</t>

<t>When a BGP speaker receives the MNH attribute on a BGP route, and re-advertises it with the nexthop altered, it processes the attribute but MUST NOT propagate it as-is. The BGP speaker MAY however attach a new instance of MNH attribute on the re-advertised route, and MAY derive its value from the received MNH. The "Advertising Router ID" field is set to the Router-ID of the speaker attaching the MNH.</t>

<t>A BGP speaker re-advertising a BGP route with nexthop unchanged MAY add the MNH attribute on the reflected BGP route, on behalf of the originating BGP speaker. The "Advertising Router ID" field in the MNH is set to the Router-ID of the speaker adding the MNH.</t>

</section>
<section anchor="interaction-with-forwarding-info-in-rest-of-update"><name>Interaction with Forwarding Info in Rest of Update</name>

<t>A type of forwarding information may be carried in both the MNH as-well as remaining portions of the Update message. E.g., AIGP, Link Bandwidth Extended Community.</t>

<t>The instance of forwarding information carried outside the MNH is associated with the BGP Nexthop (attribute codes 3, 14). The instance of forwarding information carried inside the MNH attribute is more specific, and overrides the one carried outside the MNH. It is associated with the Endpoint (<xref target="ep-tlv"/>) in Forwarding Instruction TLV (<xref target="fwd-instr-tlv"/>) that is part of same MNH attribute.</t>

<t>This rule holds good for any type of forwarding information carried in the MNH, unless specified otherwise by that type of forwarding information.</t>

<t>An exception to this rule is the MPLS Label stack information carried in <xref target="RFC8277"/> NLRI field. The Labels carried in the NLRI are imposed as inner labels with the Encapsulation Info specified in Encap Info TLV (<xref target="encap-tlv"/>) used as outer encapsulation.</t>

</section>
<section anchor="err-mnh-attr"><name>MNH attribute Error Handling</name>

<t>As specified in <xref target="RFC7606"/> BGP update message can contain no more than one instance of MP_REACH attribute or NEXT_HOP attribute. Similarly, a BGP update MUST contain no more than one instance of MNH attribute. If the MNH attribute (whether recognized or unrecognized) appears more than once in an UPDATE message, then all the occurrences of the attribute other than the first one SHALL be discarded and the UPDATE message will continue to be processed. The anomaly MAY be logged for diagnosis.</t>

<t>A TLV or sub-TLV of a certain Type in a MNH attribute can occur only once, unless specified otherwise by that type value. If multiple instances of such TLV or sub-TLV is received, the instances other than the first occurrence are ignored.</t>

<t>MNH employs a hierarchical error detection mechanism, where an error in lower layer TLVs is percolated upwards to the MNH attribute, based on the M bit. If the M bit in a TLV is 0, any error in the TLV is ignored and continue processing. If the M bit is 1, the TLV is considered invalid, and the error exception percolates up to the upper layer TLV, which takes the decision again based on it's M bit, until the MNH attribute top level. If the M bits all the way to the top were set to 1, then the error makes the MNH invalid.</t>

<t>If processing of a received MNH attribute resulted in an error, then the "M bit" is used to decide the action. If the M bit is 0, then the MNH attribute is ignored, <xref target="RFC7606"/> 'Attribute Discard' approach MUST be used, and continue to process rest of the update. If M bit is 1, then the BGP Route containing the MNH MUST be considered Unusable.</t>

<t>Implementations MAY provide policy configuration to set M bit to 0 on a MNH attribute being added, this helps with testing impact of the MNH on receiving nodes. Once confident, the MNH attribute can be re-advertised with M bit set. This helps in graceful incremental deployment.</t>

<t>The definition of a certain type of TLV or Sub-TLV in the MNH should specify in it's procedures, the value of M bit to be used. An implementation MAY provide configuration to set or reset the M bit.</t>

</section>
</section>
<section anchor="mnh-tlv"><name>MNH TLV</name>

<t>The type of MNH TLV describes how the forwarding information carried in the MNH TLV is used.</t>

<figure title="MNH TLV" anchor="fig-mnh-tlv"><artwork><![CDATA[
      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |  MNH-TLV Flags| MNH Type Code |          Length               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                              Value                            |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+



 - MNH-TLV Flags (1 octet)

           0 1 2 3 4 5 6 7
          +-+-+-+-+-+-+-+-+
          |R R R R R R R M|
          +-+-+-+-+-+-+-+-+

       All bits are reserved.

           M: "Mandatory".
              Value 1 indicates that this MNH TLV is mandatory.
              If this MNH TLV is not understood, the MNH attribute
              containing it is considered invalid.

           R: "Reserved".
              MUST be set to zero, SHOULD be ignored by receiver.

 - MNH Type Code (1 octet)
      Type of MNH TLV. 0 is Reserved.

 - Length
    Length of Value portion in octets.

 - Value
    Value portion contains the NFI TLV.

]]></artwork></figure>

<t>A sending BGP speaker advertises the information for one ore more nexthops in a MNH TLV.</t>

<t>Information received in MNH TLV is used to create the Forwarding state at receiving BGP speaker.</t>

<t>The MNH Type code indicates how the information carried in the TLV is used at the receiving node.</t>

<section anchor="err-mnh-tlv"><name>MNH TLV Error Handling</name>

<t>If invalid Type Code 0 is received, the MNH TLV is ignored irrespective of "M bit", and continue to process rest of the update.</t>

<t>If the received Type Code is incompatible for the prefix in BGP NLRI, the MNH TLV is considered invalid.</t>

<t>If an unrecognized Type Code is received, or processing of a recognized MNH TLV type results in an error, the TLV is considered invalid.</t>

<t>Invalid MNH TLV is handled based on the "M bit" on the MNH TLV.</t>

<t>If the M bit is 0, then the MNH TLV is ignored and continue to process rest of the update. If M bit is 1, then the MNH attribute containing this MNH TLV is considered invalid, triggering the procedures in <xref target="err-mnh-attr"/>.</t>

</section>
</section>
<section anchor="nexthop-forwarding-information-tlv"><name>Nexthop Forwarding Information TLV</name>

<t>A Nexthop Forwarding Information TLV describes a MNH TLV. It contains one or more Forwarding Instruction TLVs. These Forwarding Instructions are the Forwarding Legs of the MNH.</t>

<figure title="Nexthop Forwarding Information TLV" anchor="nfi-tlv"><artwork><![CDATA[
       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  NFI  Flags   |      Num-Nexthops             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |        Forwarding Instruction TLV (F.I. TLV)                  ~
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      ~        Forwarding Instruction TLV (F.I. YLV)                  |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

 - NFI Flags (1 octet)

           0 1 2 3 4 5 6 7
          +-+-+-+-+-+-+-+-+
          |R R R R R R R M|
          +-+-+-+-+-+-+-+-+

           M: "Mandatory".
              Value 1 indicates that this NFI TLV (Nexthop Leg) is mandatory.
              If this Nexthop Leg is not understood, the MNH TLV
              containing it is considered invalid.

           R: "Reserved".
              MUST be set to zero, SHOULD be ignored by receiver.

 - Num-Nexthops
        Number of F.I. TLVs.

 - Forwarding Instruction TLV
        Each F.I. TLV describes a Nexthop Leg.
        Layout of Forwarding Instruction TLV is described in next section.
]]></artwork></figure>

<t>M bit on a NFI TLV SHOULD be set to 1.</t>

<section anchor="err-nfi-tlv"><name>NFI TLV Error Handling</name>

<t>If Num-Nexthops in a received NFI is 0, it is considered invalid. Irrespective of M bit value, the NFI TLV is ignored and remaining update is processed.</t>

<t>The receiving BGP speaker MAY consider the "Num-Nexthops" value in a Nexthop Forwarding Information TLV not acceptable, based on it's forwarding capabilities or local policy. In such cases, the NFI TLV is considered Invalid.</t>

<t>An Invalid NFI TLV is handled based on value of M bit on it. If the M bit is 0, the NFI TLV is ignored, and remaining update continue to be processed. If M bit is 1, the MNH TLV carrying this NFI is considered Invalid, triggering the procedures in <xref target="err-mnh-tlv"/>.</t>

</section>
</section>
<section anchor="fwd-instr-tlv"><name>Forwarding Instruction TLV</name>

<t>Each Forwarding Instruction TLV describes a Nexthop Leg. It expresses a "Forwarding Action" (FwdAction) along with arguments required to complete the action. The type of actions defined by this TLV are given below. The arguments are denoted by "Forwarding Argument TLVs". The Forwarding Argument TLVs takes appropriate values based on the FwdAction.</t>

<t>Each FwdAction should note the Arguments needed to complete the action. Any extraneous arguments should be ignored. If the minimum set of arguments required to complete an action is not received, the Forwarding Instruction TLV should be ignored. Appropriate logging and diagnostic info MAY be provided by an implementation to help troubleshoot such scenarios.</t>

<figure title="Forwarding Instruction TLV" anchor="fig-fi-tlv"><artwork><![CDATA[
       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  F.I. Flags   |          Relative Pref        |  FwdAction    |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |            Length             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                   Fwd Argument TLV                            ~
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      ~                   Fwd Argument TLV                            |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+



  - F.I. Flags (1 octet)

           0 1 2 3 4 5 6 7
          +-+-+-+-+-+-+-+-+
          |R R R R R R R M|
          +-+-+-+-+-+-+-+-+

           M: "Mandatory".
              Value 1 indicates that this Forwarding Instruction is mandatory.
              If this instruction is not understood, the NFI TLV
              containing it is considered invalid.

           R: "Reserved".
              MUST be set to zero, SHOULD be ignored by receiver.

 - Relative Pref (2 octets)

     Unsigned 2 octet integer specifying relative order or preference, among
     the many forwarding instructions, to use in FIB. All usable nexthop legs
     with lowest relative-pref are installed in FIB as primary-path. Thus if
     multiple legs exist with that lowest relative-pref, ECMP is formed.

 - FwdAction (1 octet)

     Type Code denoting the Forwarding action to be performed by receiving node.
     0 is Reserved.

 - Length (2 octets)

    Length in octets, of all Forwarding Argument TLVs.
]]></artwork></figure>

<t>Definition of a Forwarding Action should specify the set of forwarding arguments required to execute the action, and value of M bit.</t>

<section anchor="err-fi-tlv"><name>FI TLV Error Handling</name>

<t>If an Invalid value of 0 is received as FwdAction, the TLV is ignored irrespective of "M bit", and continue to process rest of the update..</t>

<t>If an unrecognized or unsupported FwdAction is received, the FI TLV is considered Invalid.</t>

<t>If a certain Forwarding Action is unable to be executed because the set of required arguments are not available, the FI TLV is considered Invalid. If a certain Forwarding Action is applied to an incompatible NLRI, the FI TLV is considered Invalid.</t>

<t>An Invalid FI TLV is handled based on value of M bit on it. If the M bit is 0, the FI TLV is ignored, and remaining update continue to be processed. If M bit is 1, the NFI TLV carrying this NFI is considered Invalid, triggering the procedures in <xref target="err-nfi-tlv"/>.</t>

</section>
</section>
<section anchor="forwarding-argument-tlv"><name>Forwarding Argument TLV</name>

<t>The Forwarding Argument TLV describes various parameters required to execute a FwdAction.</t>

<figure title="Forwarding Argument TLV" anchor="fig-farg-tlv"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code            |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     |     Value                                     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


 - F.A. Flags (1 octet)

           0 1 2 3 4 5 6 7
          +-+-+-+-+-+-+-+-+
          |R R R R R E C M|
          +-+-+-+-+-+-+-+-+

           M: "Mandatory".
              Value 1 indicates that this argument is mandatory for the Forwarding Action.
              If this argument is not understood, the FI TLV containing it
              is considered invalid.

           C: "Cumulative".
              Request nodes to accumulate value in re-advertised MNH.
              By default Forwarding Arguments are not cumulative, so C bit is 0
              unless otherwise specified by the forwarding argument type.

           E: "Egress Attached".
              This bit is maintained when C bit is set to 1.
              E bit is set to 1 if a cumulative argument is being added to a route
              with empty AS-path.

           R: "Reserved".
              MUST be set to zero, SHOULD be ignored by receiver.


 - F.A. Type Code (2 octets)

    Type Code of Forwarding Argument. 0 is Reserved.

 - Length (2 octets)

    Length in bytes of Value field.


]]></artwork></figure>

<t>The C bit is set to 1 on attributes that need to be accumulated across BGP nexthop-self propagation hops. If a received MNH has a FA with C bit 1, it MUST be set to 1 on the FA inserted in any advertised MNH also. The value of the FA in the advertised MNH MAY be derived from the value of the FA in the received MNH. The specific FA SHOULD define the procedure on how the accumulation of value happens for the specific type of FA.</t>

<t>If a received MNH has a FA with C bit 1, and receiving speaker is unable to perform the accumulation of FA, it MUST NOT include the FA type in any advertised MNH.</t>

<t>A FA that need to be accumulated end-to-end may want to know if the cumulative value denotes the path until the Egress node. The E bit denotes that the FA was originated by the Egress node that originated this BGP route. The E bit is set to 1 by the node adding the FA, if the AS-path on the route is empty. The E bit value received on a MNH MUST be propagated on the MNH added to the re-advertisement. This allows the Ingress node to see the E bit value set by the Egress node.</t>

<section anchor="err-farg-tlv"><name>FA TLV Error Handling</name>

<t>If an Invalid F.A. Type Code value of 0 is received, the TLV is ignored irrespective of "M bit", and continue to process rest of the update..</t>

<t>If an unrecognized F.A. Type Code is received, the FA TLV is considered Invalid.</t>

<t>An Invalid FA TLV is handled based on value of M bit on it. If the M bit is 0, the FA TLV is ignored, and remaining update continue to be processed. If M bit is 1, the FI TLV carrying this FA is considered Invalid, triggering the procedures in <xref target="err-fi-tlv"/>.</t>

</section>
</section>
<section anchor="interaction-with-addpath"><name>Interaction with Addpath</name>

<t><xref target="I-D.draft-ietf-idr-add-paths-guidelines-08"/> Sec 2 suggests the following:</t>

<t>"Diverse path: A BGP path associated with a different BGP next-hop and BGP router than some other set of paths. The BGP router associated with a path is inferred from the ORIGINATOR_ID attribute or, if there is none, the BGP Identifier of the peer that advertised the path."</t>

<t>When selecting "diverse paths" for ADD_PATH as specified above, the MNH attribute should also be compared if it exists, to determine if two routes have "different BGP next-hop".</t>

</section>
<section anchor="path-selection-considerations"><name>Path Selection Considerations</name>

<section anchor="determining-igp-cost"><name>Determining IGP Cost</name>

<t>While tie breaking in the path-selection as described in <xref target="RFC4271"/>, 9.1.2.2. step (e) viz. the "IGP cost to nexthop", consider the highest cost among the nexthop-legs present in this attribute.</t>

<t>The IGP cost thus calculated is also used when constructing AIGP TLV (<xref target="RFC7311"/>)</t>

</section>
</section>
</section>
<section anchor="tlvs-defined-in-this-document"><name>TLVs Defined In This Document</name>

<t>This section describes the initial set of MNH TLVs, Forwarding Instructions and Arguments that this document defines.</t>

<section anchor="mnh-tlvs"><name>MNH TLVs</name>

<t>The type of MNH TLV describes how the forwarding information carried in the MNH TLV is used.</t>

<figure title="MNH TLV Types" anchor="fig-mnh-tlv-def"><artwork><![CDATA[
 This document defines the following MNH TLV types:

  MNH Type Code        Meaning
 --------------     -------------
       0           Reserved
       1           Primary forwarding path
       2           Repair forwarding path


 - Length
    Length of Value portion in octets.

 - Value
    Value portion contains the NFI TLV.
]]></artwork></figure>

<t>Type codes 1 and 2 are applicable for upstream allocated prefixes, example IP, Upstream MPLS labels, Flowspec routes.</t>

<t>Note that usage of Type code 1 in a BGP route containing IP prefix gives similar result as advertising the route with nexthop contained in BGP path-attributes: Nexthop (code 3) or MP_REACH_NLRI (code 14).</t>

<t>Upstream allocation for MPLS routes is achieved by using mechanisms explained in <xref target="I-D.draft-kaliraj-bess-bgp-sig-private-mpls-labels-06"/>.</t>

<t>If an invalid Type Code 0 is received, the TLV is ignored, and continue to process rest of the update.</t>

<t>If the received Type Code is incompatible for the prefix in BGP NLRI, the TLV should be ignored.</t>

<section anchor="primary-path-tlv"><name>Primary Forwarding Path</name>

<t>This is a MNH TLV (<xref target="mnh-tlv"/>) with MNH Type Code = 1, called "Primary Forwarding Path"</t>

<t>This TLV describes forwarding state to be programmed at receiving speaker as Primary Path nexthop leg. This TLV is used with Upstream allocated or global scope prefixes carried in BGP NLRI. Value part of this TLV contains Nexthop Forwarding Information TLV.</t>

<t>A BGP speaker uses the nexthop forwarding information received in this TLV as a primary path nexthop leg when programming the route for the NLRI prefix in its Forwarding table.</t>

<figure title="Primary forwarding path TLV" anchor="prim-path-tlv"><artwork><![CDATA[
      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |  MNH-TLV Flags|  MNH Type = 1 |          Length               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |               Nexthop Forwarding Information TLV              |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


]]></artwork></figure>

</section>
<section anchor="repair-path-tlv"><name>Repair Forwarding Path</name>

<t>This is a MNH TLV (<xref target="mnh-tlv"/>) with MNH Type Code = 2, called "Repair Forwarding Path"</t>

<t>This TLV describes forwarding state to be programmed during traffic repair at receiving speaker. i.e. This TLV is used to program a backup/repair path. This TLV is used with Upstream allocated prefixes or global scoped prefixes. Value part contains Nexthop Forwarding Information TLV.</t>

<t>Signaling a different nexthop for use as backup path is desirable in some labeled forwarding scenarios, where two multihomed edge devices use each other as backup path to protect traffic when primary path fails.</t>

<t>This is required to avoid label advertisement oscillation between the multihomed PEs when they implement per-nexthop label allocation mode.</t>

<t>The label advertised by a PE1 for primary path advertisement is allocated/forwarded using external paths as primary leg and backup-path label from other multihomed PE2 as backup-path label. Such that primary-path label allocation at PE1 is not a function of the primary-path label advertised by PE2. Thus the primary path label remains stable at a PE and does not change when a new primary path label is received from the other multihomed PE. This prevents the label oscillation problem.</t>

<figure title="Repair forwarding path TLV" anchor="fig-repair-path-tlv"><artwork><![CDATA[
      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |  MNH-TLV Flags|  MNH Type = 2 |          Length               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |               Nexthop Forwarding Information TLV              |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


]]></artwork></figure>

<t>The backup path label allocated and advertised by a PE is a function of only the primary path. E.g. path to the CE device. So this label value does not change when a new label is received from the other multihomed PE</t>

</section>
</section>
<section anchor="forwarding-actions-in-fi-tlv"><name>Forwarding Actions in FI TLV</name>

<t>Each Forwarding Instruction TLV describes a Nexthop Leg. It expresses a "Forwarding Action" (FwdAction) along with arguments required to complete the action. The type of actions defined by this TLV are given below. The arguments are denoted by "Forwarding Argument TLVs". The Forwarding Argument TLVs takes appropriate values based on the FwdAction.</t>

<t>Each FwdAction should note the Arguments needed to complete the action. Any extraneous arguments should be ignored. If the minimum set of arguments required to complete an action is not received, the Forwarding Instruction TLV should be ignored. Appropriate logging and diagnostic info MAY be provided by an implementation to help troubleshoot such scenarios.</t>

<t>Following Forwarding Actions are defined by this document.</t>

<figure title="Forwarding Actions" anchor="fact-def"><artwork><![CDATA[
 FwdAction         Meaning
 ---------      -------------
       0        Reserved
       1        Forward
       2        Pop-And-Forward
       3        Swap
       4        Push
       5        Pop-And-Lookup
       6        Replicate

   Forwarding Instruction TLV with unknown FwdAction should be ignored, skipped
   and rest of the attribute processed; gracefully handling the error. The event
   may be appropriately logged for diagnosis.

 - Length (2 octets)

    Length in octets, of all Forwarding Argument TLVs.
]]></artwork></figure>

<t>Meaning of most of the above FwdAction semantics is well understood. FwdAction 1 is applicable for both IP and MPLS routes. FwdActions 2-5 are applicable for encapsulated payloads (like MPLS) only. FwdActions 1, 6 are applicable for Flowspec routes for Redirect and Mirror actions. FwdAction 6 can also be used to indicate multicast replication like functionality.</t>

<t>The "Forward" action means forward the IP/MPLS packet with the destination prefix (IP-dest-addr/MPLS-label) value unchanged. For IP routes, this is the forwarding-action given for next-hop addresses contained in BGP path-attributes: Nexthop (code 3) or MP_REACH_NLRI (code 14). For MPLS routes, usage of this action is equivalent to SWAP with same label-value; one such usage is explained in <xref target="I-D.draft-kaliraj-bess-bgp-sig-private-mpls-labels-06"/> when Upstream-label-allocation is in use.</t>

<t>The "Pop-And-Forward" action means Pop the payload header (e.g. MPLS-label) and forward the payload towards the Nexthop IP-address specified in the Endpoint Id TLV, using appropriate encapsulation to reach the Nexthop.</t>

<t>When applied to MPLS packet, the "Pop-And-Lookup" action may result in a MPLS-lookup or an upper-layer header (like IPv4, IPv6) lookup, depending on whether the label that was popped was the bottom of stack label.</t>

<t>If an incompatible FwdAction is received for a prefix-type, or an unsupported FwdAction is received, it is considered a semantic-error and MUST be dealt with as explained in "Error handling procedures" section.</t>

</section>
<section anchor="farg-tlv"><name>Forwarding Argument TLVs</name>

<t>The Forwarding Argument TLV describes various parameters required to execute a FwdAction.</t>

<t>Following types of Forwarding Argument are defined by this document.</t>

<figure title="Forwarding Argument Types" anchor="farg-def"><artwork><![CDATA[
  F.A. Type Code  Meaning
  -------------  ---------
     0           Reserved
     1           Endpoint Identifier
     2           Path Constraints
     3           Payload encapsulation info signaling
     4           Endpoint attributes advertisement

 - Length (2 octets)

    Length in bytes of Value field.

]]></artwork></figure>

<section anchor="ep-tlv"><name>Endpoint Identifier</name>

<t>This is a Forwarding Argument (<xref target="farg-tlv"/>) with F.A. Type Code = 1. It identifies an Endpoint of certain type.</t>

<figure title="Endpoint Identifier" anchor="ep-def"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code =1         |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Endpoint Type |  Endpoint Len | Endpoint Value|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                  Endpoint Value                               |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


 - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

 - Length (2 octets)
    Length in bytes of Value field.


  Endpoint Type   Value                    Len (octets)
  -------------  ---------                ---------------------
     0           Reserved
     1           IPv4 Address                4
     2           IPv6 Address                16
     3           MPLS Label (Upstream        4
                            allocated or
                            Global scope)
     4           Fwd Context RD              8
     5           Fwd Context RT              8

 - Endpoint Len (1 octet)

    Length in bytes of Endpoint Value field.


]]></artwork></figure>

</section>
<section anchor="constrain-tlv"><name>Path Constraints</name>

<t>This is a Forwarding Argument (<xref target="farg-tlv"/>) with F.A. Type Code = 2. It defines Constraints for Path to the Endpoint..</t>

<figure title="Path Constraints" anchor="path-constrain"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code = 2        |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | ConstrainType | Constrain Len | ConstrainValue|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                  ConstrainValue                               |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


   - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

   - Length (2 octets)
       Length in bytes of Value field.

  ConstrainType             Value                Len (octets)
  -------------  -------------------------    ---------------------
     0           Reserved
     1           Proximity check                 2
     2           Transport Class ID (Color)      4
     3           Load balance factor             2

  - Constrain Len (1 octet)

    Length in bytes of Constrain Value field.

   - Proximity check Flags (2 octets)
        Flags describing whether the nexthop endpoint is expected to be single hop
        away, or multihop away. Format of flags is described in next section.

   - Transport Class ID (Color):

    This is a 32 bit identifier, associated with the Nexthop address.
    The Nexthop IP-address specified in "Endpoint Identifier" TLVs
    are resolved over tunnels of this color.
    Defined in {{RFC9832}}

   - Load balance factor (2 octets)
          Balance Percentage

]]></artwork></figure>

<section anchor="proximity-check"><name>Proximity Check</name>

<t>Usually EBGP singlehop received routes are expected to be one hop away, directly connected. And IBGP received routes are expected to be multihop away. Implementations today provide configuring exceptions to this rule.</t>

<t>The 'expected proximity' of the Nexthop can be signaled to the receiver using the Proximity check flags. Such that irrespective of whether the route is received from IBGP/EBGP peer, it can be treated as a single-hop away or multihop away nexthop.</t>

<t>The format of the Proximity check Sub-TLV is as follows:</t>

<figure title="Proximity check constrain" anchor="constrain-proximity"><artwork><![CDATA[
   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
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |  F.A. Flags   |     F.A. Type Code = 2        |  Length       |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |    Length     |ConstrainType=1|  Len = 2      |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |       Proximity Check Flags   |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

  - Length (2 octets)
       Length in bytes of Value field.

  - Proximity check Flags (2 octets)

           0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
          |S M R R R R R R R R R R R R R R|
          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


           S: Restrict to Singlehop path.
           M: Expect Multihop path.
           R: Reserved. MUST be set to zero, SHOULD be ignored by receiver.

]]></artwork></figure>

<t>This TLV would be valid with Forwarding Instructions TLV with FwdAction of Forward, Pop-And-Forward, Swap or Push.</t>

<t>When S bit is set, receiver considers the nexthop valid only if it is directly connected to the receiver.</t>

<t>When M bit is set, receiver assumes that the nexthop can be multiple hops away, and resolves the path to the nexthop via another route.</t>

<t>When both S and M bits are set, M bit behavior takes precedence. When both S and M bits are Clear, the current behavior of deriving proximity from peer type (EBGP is singlehop, IBGP is multihop) is followed.</t>

</section>
<section anchor="tc-tlv"><name>Transport Class ID (Color)</name>

<t>The Nexthop can be associated with a Transport Class, so as to resolve a path that satisfies required Transport tunnel characteristics. Transport Class is defined in <xref target="RFC9832"/></t>

<t>Transport Class is a per-nexthop scoped attribute. Without MNH, the Transport class is applied to the nexthop IP-address encoded in the BGP-Nexthop attribute (code 3), or inside the MP_REACH_NLRI attribute (code 14). With MNH, the Transport Class can be specified per Nexthop-Leg (Forwarding Instruction TLV <xref target="fwd-instr-tlv"/>). It is applied to the IP-address encoded in the Endpoint Identifier TLV of type "IPv4 Address", "IPv6 Address" , "MPLS Label (Upstream allocated or Global scope)".</t>

<t>The format of the Transport Class ID Sub-TLV is as follows:</t>

<figure title="Transport Class ID (Color)" anchor="constrain-tc"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code = 2        |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     |ConstrainType=2|  Len = 4      | Transport..   |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  .. Class ID (4 bytes)        |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

  - Length (2 octets)
       Length in bytes of Value field.

  - Transport Class ID (Color):
    This is a 32 bit identifier, associated with the Nexthop address.
    The Nexthop specified in Endpoint Identifier TLVs
    are resolved over tunnels of this color.
  Defined in {{RFC9832}}

]]></artwork></figure>

<t>This TLV would be valid with Forwarding Instructions TLV with FwdAction of Forward, Swap or Push.</t>

</section>
<section anchor="lb-perc"><name>Load Balance Factor</name>

<figure title="Load Balance Factor" anchor="constrain-lb"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code = 2        |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     |ConstrainType=3|  Len = 2      |   Balance..   |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|.. Percentage  |
+-+-+-+-+-+-+-+-+

 - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

 - Length (2 octets)
       Length in bytes of Value field.

 - Len (1 octet)
    Length of the Constrain Value field.

 - Balance Percentage:
    This is the explicit "balance percentage" requested by the sender,
    for unequal load-balancing over these Nexthop-Descriptor-TLV legs.
    This balance percentage would override the implicit
    balance-percentage calculated using "Bandwidth" attribute
    sub-TLV.

]]></artwork></figure>

<t>This sub-TLV would be valid with Forwarding Instructions TLV with FwdAction of Forward, Swap or Push.</t>

<t>This is the explicit "balance percentage" requested by the sender, for unequal load-balancing over these Nexthop-Descriptor-TLV legs. This balance percentage would override the implicit balance-percentage calculated using "Bandwidth" attribute sub-TLV</t>

<t>When the sum of "balance percentage" on the nexthop legs does not equal 100, it is scaled up or down to match 100. The individual balance percentages in each nexthop leg are also scaled up or down proportionally to determine the effective balance percentage per nexthop leg.</t>

</section>
</section>
<section anchor="encap-tlv"><name>Payload Encapsulation Info</name>

<t>This is a Forwarding Argument (<xref target="farg-tlv"/>) with F.A. Type Code = 3. It defines Payload Encapsulation Information.</t>

<figure title="Payload Encapsulation Info" anchor="encap-info"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code =3         |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Encap Type  |         Encap Len               |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                   Encap Value                                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


 - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

 - Length (2 octets)
       Length in bytes of Value field.

   Endcap Type        Value
  -------------  --------------
     0           Reserved
     1           MPLS Label Info
     2           SR MPLS label Index Info
     3           SRv6 SID info
     4           DSCP code point

 - Encap Len (2 octets)

    Length in octets of Encap Value field.


]]></artwork></figure>

<section anchor="mpls-label-info"><name>MPLS Label Info</name>

<figure title="MPLS Label Info" anchor="encap-mpls"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code =3         |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Encap Type=1 |          Encap Len             |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|      L.V. Flags (2 bytes)    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| MPLS Label (20 bits) |Rsrv |S~
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
~ MPLS Label (20 bits) |Rsrv |S|
-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


  - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

  - Length (2 octets)
       Length in bytes of Value field.

  - Encap Type
     = 1, to signify MPLS Label Info.

  - Encap Len (2 octets)
       Length in bytes of following Encap Value field.

  - L.V. Flags (2 octets):

       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |E R R R R R R R R R R R R R R R|
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

       E: ELC bit. Indicates if this egress NH is Entropy Label Capable.
             1 means the Entropy Label capable.
             0 means not capable to handle Entropy Label.

       R: Reserved. MUST be set to zero, SHOULD be ignored by receiver.

  - MPLS Label, Rsrv, S bit.
      20 bit MPLS Label stack encoded as in RFC 8277.
      S bit set on last label in label stack.


]]></artwork></figure>

</section>
<section anchor="sr-mpls-label-index-info"><name>SR MPLS Label Index Info</name>

<figure title="SR MPLS Label Index Info" anchor="encap-sr-mpls"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code =3         |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Encap Type=2 |            Encap Len           |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|   RESERVED    |       LI Flags                |    Label ..   |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                ..Index                        |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


  - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

  - Length (2 octets)
       Length in bytes of Value field.

  - Encap Type
     = 2, to signify SR MPLS SID Info.

  - Encap Len (2 octets)
       Length in bytes of following Encap Value field.

  Rest of the value portion is encoded as specified in RFC-8669 sec 3.1.

  - RESERVED:  8-bit field. MUST be set to zero, SHOULD be ignored by receiver.

  - LI Flags:  16 bits of flags. None defined. MUST be set to zero, SHOULD be ignored by receiver.

  - Label Index:
      32-bit value representing the index value in the SRGB space.

]]></artwork></figure>

</section>
<section anchor="srv6-sid-info"><name>SRv6 SID Info</name>

<figure title="SRv6 SID Info" anchor="encap-srv6"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code =3         |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Encap Type=3 |           Encap Len            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|         .. SRv6 SID Info (variable)                           |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


  - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

  - Length (2 octets)
       Length in bytes of Value field.

  - Encap Type
        = 3, to signify SRv6 SID Info.

  - Encap Len (2 octets)
       Length in bytes of following Encap Value field.

  - SRv6 SID Info:
        SRv6 SID Information, as specified in RFC-9252 sec 3.1.


]]></artwork></figure>

</section>
<section anchor="dscp"><name>DSCP</name>

<figure title="DSCP" anchor="encap-dscp"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code = 3        |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Encap Type=4 |           Encap Len            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|DSCP code point|
+-+-+-+-+-+-+-+-+


  - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

  - Length (2 octets)
       Length in bytes of Value field.

  - Encap Type
        = 4, to signify DSCP code point.

  - Encap Len (2 octets)
      = 1, Length in bytes of following Encap Value field.

  - DSCP code point:
        DS Field, as specified in RFC-2474 sec 3.


]]></artwork></figure>

</section>
</section>
<section anchor="epattr-tlv"><name>Endpoint Attributes</name>

<t>This is a Forwarding Argument (<xref target="farg-tlv"/>) with F.A. Type Code = 4. It defines Attributes of an Endpoint.</t>

<figure title="Endpoint attributes" anchor="ep-attr"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code = 4        |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Attrib Type  |    Attr Len    |  Attr  Value  |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                    Attr Value                                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


   EP Attrib Type      Attrib Value               Attrib Len (octets)
  ----------------  ------------------            ---------------------
     0               None
     1               Endpoint Bandwidth               8
     2               Accumulated Metric               Variable


]]></artwork></figure>

<section anchor="ep-bw"><name>Endpoint Bandwidth</name>

<figure title="Endpoint Bandwidth" anchor="ep-attr-bw"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code = 4        |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Attrib Type 1|    Attr Len=8  |  Attr  Value  |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                   Bandwidth (8 octets)                        |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                   Bandwidth (contd.)                          |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

 - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

 - Len (2 octets)
    Length in bytes of remaining portion of SubTLV.

 - Bandwidth
    The bandwidth to the endpoint expressed as 8 octets,
    units being bits per second.

]]></artwork></figure>

<t>This sub-TLV would be valid with Forwarding Instruction TLV with FwdAction of Forward, Swap or Push.</t>

</section>
<section anchor="accumulated-metric-to-endpoint"><name>Accumulated Metric to Endpoint</name>

<figure title="Accumulated Metric to Endpoint" anchor="ep-attr-acc-metric"><artwork><![CDATA[
 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|  F.A. Flags   |     F.A. Type Code = 4        |  Length       |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|    Length     | Attrib Type 2|    Metric Type |  Metric Len   |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                   Metric Value (variable)                     |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


 - F.A. Flags (1 octet)
     As defined in Forwarding Argument TLV.

        C: Cumulative bit is set to 1 by originator of this argument.


- Len (2 octets)
    Length in bytes of remaining portion of SubTLV.

- Metric Type: Type from "IGP Metric-Type" IANA registry under IGP Parameters
   Following types are defined by this document to be accumulated:
      0 IGP Metric
      1 Min Unidirectional Link Delay as defined in [RFC8570, Section 4.2]

- Metric Len: Length in octets of Metric Value field.
      IGP Metric: 4
      Min Unidirectional Link Delay: 4

- Metric Value:
      IGP Metric: 4 octet Accumulated IGP cost
      Min Unidirectional Link Delay: 4 octet Accumulated min delay in microseconds.

]]></artwork></figure>

<t>This sub-TLV would be valid with Forwarding Instruction TLV with FwdAction of Forward, Swap or Push.</t>

</section>
</section>
</section>
</section>
<section anchor="scaling-considerations"><name>Scaling Considerations</name>

<t>The MNH attribute allows receiving multiple nexthops on the same BGP session. This flexibility also opens up the possibility that a peer can send large number of multipath (ECMP/UCMP/FRR) nexthops that may overwhelm the local system's forwarding plane. Prefix-limit based checks will not avoid this situation.</t>

<t>To keep the scaling limits under check, a BGP speaker MAY keep account of number of unique multipath nexthops that are received from a BGP peer, and impose a configurable max-limit on that. This is especially useful for EBGP peers.</t>

<t>A good scaling property of conveying multipath nexthops using the MNH attribute with N nexthop legs on one BGP session, as against BGP routes on N BGP sessions is that, it limits the amount of transitionary multipath combinatorial state in the latter model. Because the final multipath state is conveyed by one route update in deterministic manner, there is no transitionary multipath combinatorial explosion created during establishment of N sessions.</t>

</section>
<section anchor="iana-considerations"><name>IANA Considerations</name>

<t>This document makes request to IANA to allocate the following codes in BGP attributes registry.</t>

<section anchor="bgp-path-attributes"><name>BGP Path Attributes</name>

<t>A new BGP attribute code TBD for "BGP MultiNexthop Attribute (MULTI_NEXT_HOP)", in "BGP Path Attributes" registry.</t>

</section>
<section anchor="bgp-multinexthop-attribute"><name>BGP MultiNextHop Attribute</name>

<t>This document requests IANA to create a new registry group for MultiNextHop attribute, and the following registries in it.</t>

<section anchor="multinexthop-mnh-tlv-types"><name>MultiNextHop (MNH) TLV Types</name>

<t>This is a Registry for Type codes in (<xref target="mnh-tlv"/>) "MNH TLV"</t>

<figure title="Registry - MNH TLV Types" anchor="reg-mnh-tlv"><artwork><![CDATA[
   Under "Border Gateway Protocol (BGP) Parameters",

     Registry Group: BGP MultiNextHop Attribute

     Registry Name: MultiNexthop (MNH) TLV Types

          MNH Type Code        Meaning
         --------------     -------------
           0                Reserved
           1                Primary forwarding path
           2                Repair forwarding path
           3-254            Unassigned
           255              Reserved

     Reference: This document.

     Registration Procedure(s)
         Future assignments are to be made using either the Standards Action
         process defined in [RFC2434], or the Early IANA Allocation process
         defined in [RFC4020].


]]></artwork></figure>

</section>
<section anchor="forwarding-action-types"><name>Forwarding Action Types</name>

<t>This is a Registry for Type codes in <xref target="fwd-instr-tlv"/> "Forwarding Instruction TLV"</t>

<figure title="Registry - Forwarding Action Types" anchor="reg-fact"><artwork><![CDATA[
    Under "Border Gateway Protocol (BGP) Parameters",

      Registry Group: BGP MultiNextHop Attribute

      Registry Name: Forwarding Action Types

            FwdAction         Meaning
            ---------      -------------
             0             Reserved
             1             Forward
             2             Pop-And-Forward
             3             Swap
             4             Push
             5             Pop-And-Lookup
             6             Replicate
             7-254         Unassigned
             255           Reserved

       Reference: This document.

       Registration Procedure(s)
           Future assignments are to be made using either the Standards Action
           process defined in [RFC2434], or the Early IANA Allocation process
           defined in [RFC4020].

]]></artwork></figure>

</section>
<section anchor="forwarding-argument-types"><name>Forwarding Argument Types</name>

<t>This is a Registry for Type codes in (<xref target="farg-tlv"/>) "Forwarding Arguments TLV"</t>

<figure title="Registry - Forwarding Argument Types" anchor="reg-farg"><artwork><![CDATA[
    Under "Border Gateway Protocol (BGP) Parameters",

      Registry Group: BGP MultiNextHop Attribute

      Registry Name: Forwarding Argument Types

         F.A. Type Code      Meaning
         ---------------   ------------------
            0              Reserved
            1              Endpoint Identifier
            2              Path Constraints
            3              Payload encapsulation info signaling
            4              Endpoint attributes advertisement
            5-65534        Unassigned
            65535          Reserved

      Reference: This document.

      Registration Procedure(s)
          Future assignments are to be made using either the Standards Action
          process defined in [RFC2434], or the Early IANA Allocation process
          defined in [RFC4020].


]]></artwork></figure>

</section>
<section anchor="endpoint-types"><name>Endpoint Types</name>

<t>This is a Registry for Type codes in <xref target="ep-tlv"/> "Endpoint Identifier" Forwarding Argument.</t>

<figure title="Registry - Endpoint Types" anchor="reg-ep"><artwork><![CDATA[
    Under "Border Gateway Protocol (BGP) Parameters",

      Registry Group: BGP MultiNextHop Attribute

      Registry Name: Endpoint Types

          Endpoint Type   Value
         -------------  ---------
            0           Reserved
            1           IPv4 Address
            2           IPv6 Address
            3           MPLS Label
            4           Fwd Context RD
            5           Fwd Context RT
            6-254       Unassigned
            255         Reserved

      Reference: This document.

      Registration Procedure(s)
          Future assignments are to be made using either the Standards Action
          process defined in [RFC2434], or the Early IANA Allocation process
          defined in [RFC4020].


]]></artwork></figure>

</section>
<section anchor="path-constrain-types"><name>Path Constrain Types</name>

<t>This is a Registry for Type codes in <xref target="constrain-tlv"/> "Path Constrain" Forwarding Argument.</t>

<figure title="Registry - Path Constrain Types" anchor="reg-constrains"><artwork><![CDATA[
    Under "Border Gateway Protocol (BGP) Parameters",

      Registry Group: BGP MultiNextHop Attribute

      Registry Name: Path Constrain Types

         ConstrainType             Value
         -------------  -------------------------
           0             Reserved
           1             Proximity check
           2             Transport Class ID (Color)
           3             Load balance factor
           4-254         Unassigned
           255           Reserved

      Reference: This document.

      Registration Procedure(s)
           Future assignments are to be made using either the Standards Action
           process defined in [RFC2434], or the Early IANA Allocation process
           defined in [RFC4020].


]]></artwork></figure>

</section>
<section anchor="encapsulation-types"><name>Encapsulation Types</name>

<t>This is a Registry for Type codes in <xref target="encap-tlv"/> "Payload Encapsulation Info" Forwarding Argument.</t>

<figure title="Registry - Encapsulation Types" anchor="reg-encap"><artwork><![CDATA[
      Under "Border Gateway Protocol (BGP) Parameters",

        Registry Group: BGP MultiNextHop Attribute

        Registry Name: Encapsulation Types

            Encap Type        Value
          -------------  --------------
            0           Reserved
            1           MPLS Label Info
            2           SR MPLS label Index Info
            3           SRv6 SID info
            4           DSCP code point
            5-254       Unassigned
            255         Reserved

        Reference: This document.

        Registration Procedure(s)
            Future assignments are to be made using either the Standards Action
            process defined in [RFC2434], or the Early IANA Allocation process
            defined in [RFC4020].


]]></artwork></figure>

</section>
<section anchor="endpoint-attribute-types"><name>Endpoint Attribute Types</name>

<t>This is a Registry for Type codes in <xref target="epattr-tlv"/> "Endpoint attributes" Forwarding Argument.</t>

<figure title="Registry - Endpoint Attribute Types" anchor="reg-ep-attr"><artwork><![CDATA[
    Under "Border Gateway Protocol (BGP) Parameters",

      Registry Group: BGP MultiNextHop Attribute

      Registry Name:  Endpoint Attribute Types

         EP Attrib Type      Attrib Value
         ----------------  ------------------
           0               Reserved
           1               Bandwidth
           2               Accumulated Metric to Endpoint
           3-254           Unassigned
           255             Reserved

       Reference: This document.

       Registration Procedure(s)
           Future assignments are to be made using either the Standards Action
           process defined in [RFC2434], or the Early IANA Allocation process
           defined in [RFC4020].


]]></artwork></figure>

<t>Note to RFC Editor: this section may be removed on publication as an RFC.</t>

</section>
</section>
</section>
<section anchor="security-considerations"><name>Security Considerations</name>

<t>The MNH attribute is defined as optional non-transitive BGP attribute, such that it does not accidentally get propagated or leaked via BGP speakers that don't support this feature, especially does not unintentionally leak across EBGP boundaries.</t>

<t>MNH may be used to advertise nexthop with MPLS label in various BGP families. In scenarios where MPLS is enabled on link to a device in an untrusted domain, e.g. a PE-CE link or ASBR-ASBR inter-AS link, security can be provided against MPLS label spoofing by using MPLS context tables as described in <xref target="mpls-enabled-ce"/>. Such that only MPLS traffic with labels advertised to the BGP speaker are allowed to forward. However, the PE may not be able to perform any checks based on inner payload in the MPLS packet since it performs label swap forwarding. Such 'inner payload' based checks may be offloaded to a downstream node that forwards and processes inner payload, e.g., an IP router having full FIB. These security aspects should be considered when using MPLS enabled CE devices.</t>

</section>
<section anchor="acknowledgements"><name>Acknowledgements</name>

<t>Thanks to Jeff Haas, Robert Raszuk, Ron Bonica for the review, discussions and input to the draft.</t>

<t>Thanks to Blaine Williams and Satya Mohanty for the discussions on some use-cases.</t>

</section>


  </middle>

  <back>


<references title='References' anchor="sec-combined-references">

    <references title='Normative References' anchor="sec-normative-references">



<reference anchor="RFC7606">
  <front>
    <title>Revised Error Handling for BGP UPDATE Messages</title>
    <author fullname="E. Chen" initials="E." role="editor" surname="Chen"/>
    <author fullname="J. Scudder" initials="J." role="editor" surname="Scudder"/>
    <author fullname="P. Mohapatra" initials="P." surname="Mohapatra"/>
    <author fullname="K. Patel" initials="K." surname="Patel"/>
    <date month="August" year="2015"/>
    <abstract>
      <t>According to the base BGP specification, a BGP speaker that receives an UPDATE message containing a malformed attribute is required to reset the session over which the offending attribute was received. This behavior is undesirable because a session reset would impact not only routes with the offending attribute but also other valid routes exchanged over the session. This document partially revises the error handling for UPDATE messages and provides guidelines for the authors of documents defining new attributes. Finally, it revises the error handling procedures for a number of existing attributes.</t>
      <t>This document updates error handling for RFCs 1997, 4271, 4360, 4456, 4760, 5543, 5701, and 6368.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="7606"/>
  <seriesInfo name="DOI" value="10.17487/RFC7606"/>
</reference>
<reference anchor="RFC8277">
  <front>
    <title>Using BGP to Bind MPLS Labels to Address Prefixes</title>
    <author fullname="E. Rosen" initials="E." surname="Rosen"/>
    <date month="October" year="2017"/>
    <abstract>
      <t>This document specifies a set of procedures for using BGP to advertise that a specified router has bound a specified MPLS label (or a specified sequence of MPLS labels organized as a contiguous part of a label stack) to a specified address prefix. This can be done by sending a BGP UPDATE message whose Network Layer Reachability Information field contains both the prefix and the MPLS label(s) and whose Next Hop field identifies the node at which said prefix is bound to said label(s). This document obsoletes RFC 3107.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="8277"/>
  <seriesInfo name="DOI" value="10.17487/RFC8277"/>
</reference>
<reference anchor="RFC7311">
  <front>
    <title>The Accumulated IGP Metric Attribute for BGP</title>
    <author fullname="P. Mohapatra" initials="P." surname="Mohapatra"/>
    <author fullname="R. Fernando" initials="R." surname="Fernando"/>
    <author fullname="E. Rosen" initials="E." surname="Rosen"/>
    <author fullname="J. Uttaro" initials="J." surname="Uttaro"/>
    <date month="August" year="2014"/>
    <abstract>
      <t>Routing protocols that have been designed to run within a single administrative domain (IGPs) generally do so by assigning a metric to each link and then choosing, as the installed path between two nodes, the path for which the total distance (sum of the metric of each link along the path) is minimized. BGP, designed to provide routing over a large number of independent administrative domains (autonomous systems), does not make its path-selection decisions through the use of a metric. It is generally recognized that any attempt to do so would incur significant scalability problems as well as inter-administration coordination problems. However, there are deployments in which a single administration runs several contiguous BGP networks. In such cases, it can be desirable, within that single administrative domain, for BGP to select paths based on a metric, just as an IGP would do. The purpose of this document is to provide a specification for doing so.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="7311"/>
  <seriesInfo name="DOI" value="10.17487/RFC7311"/>
</reference>
<reference anchor="RFC4271">
  <front>
    <title>A Border Gateway Protocol 4 (BGP-4)</title>
    <author fullname="Y. Rekhter" initials="Y." role="editor" surname="Rekhter"/>
    <author fullname="T. Li" initials="T." role="editor" surname="Li"/>
    <author fullname="S. Hares" initials="S." role="editor" surname="Hares"/>
    <date month="January" year="2006"/>
    <abstract>
      <t>This document discusses the Border Gateway Protocol (BGP), which is an inter-Autonomous System routing protocol.</t>
      <t>The primary function of a BGP speaking system is to exchange network reachability information with other BGP systems. This network reachability information includes information on the list of Autonomous Systems (ASes) that reachability information traverses. This information is sufficient for constructing a graph of AS connectivity for this reachability from which routing loops may be pruned, and, at the AS level, some policy decisions may be enforced.</t>
      <t>BGP-4 provides a set of mechanisms for supporting Classless Inter-Domain Routing (CIDR). These mechanisms include support for advertising a set of destinations as an IP prefix, and eliminating the concept of network "class" within BGP. BGP-4 also introduces mechanisms that allow aggregation of routes, including aggregation of AS paths.</t>
      <t>This document obsoletes RFC 1771. [STANDARDS-TRACK]</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="4271"/>
  <seriesInfo name="DOI" value="10.17487/RFC4271"/>
</reference>
<reference anchor="RFC7911">
  <front>
    <title>Advertisement of Multiple Paths in BGP</title>
    <author fullname="D. Walton" initials="D." surname="Walton"/>
    <author fullname="A. Retana" initials="A." surname="Retana"/>
    <author fullname="E. Chen" initials="E." surname="Chen"/>
    <author fullname="J. Scudder" initials="J." surname="Scudder"/>
    <date month="July" year="2016"/>
    <abstract>
      <t>This document defines a BGP extension that allows the advertisement of multiple paths for the same address prefix without the new paths implicitly replacing any previous ones. The essence of the extension is that each path is identified by a Path Identifier in addition to the address prefix.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="7911"/>
  <seriesInfo name="DOI" value="10.17487/RFC7911"/>
</reference>



    </references>

    <references title='Informative References' anchor="sec-informative-references">



<reference anchor="RFC2474">
  <front>
    <title>Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers</title>
    <author fullname="K. Nichols" initials="K." surname="Nichols"/>
    <author fullname="S. Blake" initials="S." surname="Blake"/>
    <author fullname="F. Baker" initials="F." surname="Baker"/>
    <author fullname="D. Black" initials="D." surname="Black"/>
    <date month="December" year="1998"/>
    <abstract>
      <t>This document defines the IP header field, called the DS (for differentiated services) field. [STANDARDS-TRACK]</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="2474"/>
  <seriesInfo name="DOI" value="10.17487/RFC2474"/>
</reference>
<reference anchor="RFC9832">
  <front>
    <title>BGP Classful Transport Planes</title>
    <author fullname="K. Vairavakkalai" initials="K." role="editor" surname="Vairavakkalai"/>
    <author fullname="N. Venkataraman" initials="N." role="editor" surname="Venkataraman"/>
    <date month="September" year="2025"/>
    <abstract>
      <t>This document specifies a mechanism referred to as "Intent-Driven Service Mapping". The mechanism uses BGP to express Intent-based association of overlay routes with underlay routes having specific Traffic Engineering (TE) characteristics satisfying a certain Service Level Agreement (SLA). This is achieved by defining new constructs to group underlay routes with sufficiently similar TE characteristics into identifiable classes (called "Transport Classes" or "TCs"), that overlay routes use as an ordered set to resolve reachability (Resolution Schemes) towards service endpoints. These constructs can be used, for example, to realize the "IETF Network Slice" defined in the TEAS Network Slices framework (RFC 9543).</t>
      <t>Additionally, this document specifies protocol procedures for BGP that enable dissemination of service mapping information in a network that may span multiple cooperating administrative domains. These domains may be administered either by the same provider or by closely coordinating providers. A new BGP address family that leverages the procedures described in RFC 4364 ("BGP/MPLS IP Virtual Private Networks (VPNs)") and follows the NLRI encoding described in RFC 8277 ("Using BGP to Bind MPLS Labels to Address Prefixes") is defined to enable each advertised underlay route to be identified by its class. This new address family is called "BGP Classful Transport" (or "BGP CT").</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9832"/>
  <seriesInfo name="DOI" value="10.17487/RFC9832"/>
</reference>
<reference anchor="RFC9256">
  <front>
    <title>Segment Routing Policy Architecture</title>
    <author fullname="C. Filsfils" initials="C." surname="Filsfils"/>
    <author fullname="K. Talaulikar" initials="K." role="editor" surname="Talaulikar"/>
    <author fullname="D. Voyer" initials="D." surname="Voyer"/>
    <author fullname="A. Bogdanov" initials="A." surname="Bogdanov"/>
    <author fullname="P. Mattes" initials="P." surname="Mattes"/>
    <date month="July" year="2022"/>
    <abstract>
      <t>Segment Routing (SR) allows a node to steer a packet flow along any path. Intermediate per-path states are eliminated thanks to source routing. SR Policy is an ordered list of segments (i.e., instructions) that represent a source-routed policy. Packet flows are steered into an SR Policy on a node where it is instantiated called a headend node. The packets steered into an SR Policy carry an ordered list of segments associated with that SR Policy.</t>
      <t>This document updates RFC 8402 as it details the concepts of SR Policy and steering into an SR Policy.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9256"/>
  <seriesInfo name="DOI" value="10.17487/RFC9256"/>
</reference>

<reference anchor="I-D.draft-kaliraj-bess-bgp-sig-private-mpls-labels-06">
   <front>
      <title>BGP Signaled MPLS Namespaces</title>
      <author fullname="Kaliraj Vairavakkalai" initials="K." surname="Vairavakkalai">
         <organization>Juniper Networks, Inc.</organization>
      </author>
      <author fullname="Jeyananth Minto Jeganathan" initials="J. M." surname="Jeganathan">
         <organization>Juniper Networks, Inc.</organization>
      </author>
      <author fullname="Praveen Ramadenu" initials="P." surname="Ramadenu">
         <organization>AT&amp;T Services, Inc.</organization>
      </author>
      <date day="10" month="July" year="2023"/>
      <abstract>
	 <t>   The MPLS forwarding layer in a core network is a shared resource.
   The MPLS FIB at nodes in this layer contains labels that are
   dynamically allocated and locally significant at that node.  These
   labels are scoped in context of the global loopback address.  Let us
   call this the global MPLS namespace.

   For some usecases like upstream label allocation, it is useful to
   create private MPLS namespaces (virtual MPLS FIB) over this shared
   MPLS forwarding layer.  This allows installing deterministic label
   values in the private FIBs created at nodes participating in the
   private MPLS namespace, while preserving the &quot;locally significant&quot;
   nature of the underlying shared global MPLS FIB.

   This document defines new address families (AFI: 16399, SAFI: 128, or
   1) and associated signaling mechanisms to create and use MPLS
   forwarding contexts in a network.  Some example use cases are also
   described.


	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-kaliraj-bess-bgp-sig-private-mpls-labels-06"/>
   
</reference>

<reference anchor="I-D.draft-ietf-idr-flowspec-redirect-ip">
   <front>
      <title>BGP Flow-Spec Redirect-to-IP Action</title>
      <author fullname="Jeff Haas" initials="J." surname="Haas">
         <organization>HPE</organization>
      </author>
      <author fullname="Wim Henderickx" initials="W." surname="Henderickx">
         <organization>Nokia</organization>
      </author>
      <author fullname="Adam Simpson" initials="A." surname="Simpson">
         <organization>Nokia</organization>
      </author>
      <date day="21" month="May" year="2026"/>
      <abstract>
	 <t>   Flow-spec is an extension to BGP that allows for the dissemination of
   traffic flow specification rules.  This has many possible
   applications, but the primary one for many network operators is the
   distribution of traffic filtering actions for distributed denial of
   service (DDoS) mitigation.  The flow-spec standard, RFC 8955, defines
   a redirect-to-VRF (Virtual Routing and Forwarding) action for policy-
   based forwarding.  This mechanism can be difficult to use,
   particularly in networks without Layer 3 VPN infrastructure.

   This document defines a new redirect-to-IP flow-spec action that
   provides a simpler method of policy-based forwarding.  The details of
   the action, including the IPv4 or IPv6 target address, are encoded in
   newly defined BGP extended communities [RFC4360].

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-idr-flowspec-redirect-ip-16"/>
   
</reference>

<reference anchor="I-D.draft-ietf-idr-add-paths-guidelines-08">
   <front>
      <title>Best Practices for Advertisement of Multiple Paths in IBGP</title>
      <author fullname="Jim Uttaro" initials="J." surname="Uttaro">
         <organization>AT&amp;T</organization>
      </author>
      <author fullname="Pierre Francois" initials="P." surname="Francois">
         <organization>IMDEA Networks</organization>
      </author>
      <author fullname="Keyur Patel" initials="K." surname="Patel">
         <organization>Cisco Systems</organization>
      </author>
      <author fullname="Jeffrey Haas" initials="J." surname="Haas">
         <organization>Juniper Networks</organization>
      </author>
      <author fullname="Adam Simpson" initials="A." surname="Simpson">
         <organization>Nokia</organization>
      </author>
      <author fullname="Roberto Fragassi" initials="R." surname="Fragassi">
         <organization>Nokia</organization>
      </author>
      <date day="25" month="April" year="2016"/>
      <abstract>
	 <t>   Add-Paths is a BGP enhancement that allows a BGP router to advertise
   multiple distinct paths for the same prefix/NLRI. This provides a
   number of potential benefits, including reduced routing churn, faster
   convergence and better loadsharing.

   This document provides recommendations to implementers of Add-Paths
   so that network operators have the tools needed to address their
   specific applications and to manage the scalability impact of Add-
   Paths. A router implementing Add-Paths may learn many paths for a
   prefix and must decide which of these to advertise to peers. This
   document analyses different algorithms for making this selection and
   provides recommendations based on the target application.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-idr-add-paths-guidelines-08"/>
   
</reference>



    </references>

</references>


<?line 1543?>

<section anchor="use"><name>Use Cases</name>

<t>This section describes various example use-cases of the MNH attribute.</t>

<section anchor="signaling-wecmp-to-ingress-node"><name>Signaling WECMP to Ingress Node</name>

<t>This section describes how MNH can be used to provide weighted equal cost multipath in a network fabric, while not increasing RIB scale.</t>

<figure title="WECMP without increasing RIB scale" anchor="topo-wecmp"><artwork><![CDATA[
                                   [RR1]
                                     .
                   . +-[P21]         |
                  .  +-[P22]         __
                 .   +-[P23]      _.(  )..
            [R1].    +-[P24] ..  (_      _) .. [R2]
                 .   +-[P25]       (._..)
                  .
                   . +-[P2n]


                    <---- Traffic Direction ----

]]></artwork></figure>

<t>{topo-wecmp} shows a network with BGP speaker R1 connected to a number of routers P21 .. P2n in its region. R1 is eSN and R2 is iSN for the IP traffic in consideration. BGP service families IPv4 Unicast (AFI/SAFI: 1/1) and IPv6 Unicast (AFI/SAFI: 2/1) are negotiated on the BGP sessions between RR1 - R1 and RR1 - R2. RR1 reflects the BGP routes between R1 and R2 with next hop unchanged.</t>

<t>When MNH is not in use, R1 advertises "n" BGP Addpath routes for a service prefix Pfx1, each having a distinct next hop, P21 .. P2n, and desired Link Bandwidth Extended Community. These Addpath routes will be received by R2, which can do WECMP based on the Link Bandwidth Extended Communities attached on the routes. This model increases RIB scale by "n" times, so that WECMP can be achieved.</t>

<t>When MNH is used in this network, R1 advertises a single BGP route for prefix Pfx1, which contains a MNH attribute with "n" next hops, each carrying the desired link bandwidth using <xref target="lb-perc"/> or <xref target="ep-bw"/></t>

<t>This allows achieving WECMP in the network without increasing RIB scale.</t>

</section>
<section anchor="signaling-optimal-forwarding-exit-points-to-ingress-node"><name>Signaling Optimal Forwarding Exit-points to Ingress Node</name>

<t>In a BGP free core, one can dynamically signal to the ingress-node, how traffic should be load-balanced towards a set of exit nodes, in one BGP-route containing this attribute.</t>

<t>Example, for prefix1, perform equal load balancing towards exit nodes A, B; where as for prefix2, perform weighted load balancing (40%, 30%, 30%) towards exit nodes A, B, C.</t>

<t>Example, for prefix1, use PE1 as primary-nexthop and use PE2 as a backup-nexthop.</t>

</section>
<section anchor="load-balancing-to-multiple-ces-in-a-vrf"><name>Load balancing to multiple CEs in a VRF</name>

<t>This section describes how MNH can be used to provide load balancing and entropy in a provider network for traffic destined to multiple CEs in a VRF, without increasing RIB scale.</t>

<figure title="Load balancing to multiple CEs in a VRF" anchor="topo-cevrf"><artwork><![CDATA[

                                   [RR1]
                                     .
            [CE1].                   |
                  .                  __
            [CE2]  .              _.(  )..
                    .[PE1]   ..  (_      _) .. [PE2]
            [CE3]  .               (._..)
                  .
            [CE4].


                    <---- Traffic Direction ----
]]></artwork></figure>

<t>{topo-cevrf} shows a L3VPN network with multiple CE devices connected to the same VRF at PE1. The VRF is configured with a RD: RD1, and uses "per next hop" label allocation mode to advertise the CE routes to L3VPN core. PE1 is eSN and PE2 is iSN for the IP traffic in consideration. CE1..CE4 advertise route for same prefix Pfx1 in BGP service families IPv4 Unicast (AFI/SAFI: 1/1) negotiated on the BGP sessions between the CEs and PE1. BGP L3VPN address family (AFI/SAFI: 1/128) is negotiated between PE1 - RR1, and RR1 - PE2. RR1 reflects the BGP routes between PE1 and PE2 with next hop unchanged.</t>

<t>PE1 would typically advertise to RR1 only the best path for prefix Pfx1 out of routes received from CE1..CE4. Using per CE RD or Addpath for L3VPN family may allow PE1 to advertise all CE routes to the RR, with an increase in RIB scale. This model increases RIB scale by "n" times, where 'n' is the number of CEs.</t>

<t>When MNH is used in this network, PE1 advertises a single BGP L3VPN route for prefix Pfx1, which contains a MNH attribute with "n" next hops, each carrying the label pointing towards a particular CE, using Payload Encapsulation Info TLV (<xref target="encap-tlv"/>) along with the Endpoint Identifier TLV (<xref target="ep-tlv"/>)</t>

<t>This allows the network to direct traffic to a specific CE, and better load-balance traffic in the provider network, with entropy provided by the per CE VPN labels, without increasing RIB scale.</t>

</section>
<section anchor="signaling-desired-forwarding-behavior-for-mpls-upstream-labels-at-receiving-node"><name>Signaling Desired Forwarding Behavior for MPLS Upstream labels at Receiving Node</name>

<t>In Upstream label allocation case, the receiving speaker's forwarding-state can be controlled by the advertising speaker, thus enabling a standardized API to program desired MPLS forwarding-state at the receiving node. This is described in the <xref target="I-D.draft-kaliraj-bess-bgp-sig-private-mpls-labels-06"/></t>

</section>
<section anchor="load-balancing-over-ebgp-parallel-links"><name>Load Balancing over EBGP Parallel Links</name>

<t>Consider N parallel links between two EBGP speakers. There are different models possible to do load balancing over these links:N single-hop EBGP sessions over the N links. Interface addresses are used as next-hops. N copies of the RIB are exchanged to form N-way ECMP paths. The routes advertised on the N sessions can be attached with Link bandwidth community to perform weighted ECMP. 1 multi-hop EBGP session between loopback addresses, reachable via static route over the N links. Loopback addresses are used as next-hops. 1 copy of the RIB is exchanged with loopback address as nexthop. And a static route can be configured to the loopback address to perform desired N-way ECMP path. M loopbacks are configured in this model, to achieve M different load balancing schemes: ECMP, weighted ECMP, Fast-reroute enabled paths etc. 1 multi-hop EBGP session between loopback addresses, reachable via static route over the N links. Interface addresses are used as next-hops, without using additional loopbacks. 1 copy of the RIB is exchanged with MNH attribute to form N-way ECMP paths, weighted ECMP, Fast-reroute backup paths etc. BFD may be used to these directly connected BGP nexthops to detect liveness.</t>

<t><list style="symbols">
  <t>N single-hop EBGP sessions over the N links. Interface addresses are used as next-hops. N copies of the RIB are exchanged to form N-way ECMP paths. The routes advertised on the N sessions can be attached with Link bandwidth community to perform weighted ECMP.</t>
  <t>1 multi-hop EBGP session between loopback addresses, reachable via static route over the N links. Loopback addresses are used as next-hops. 1 copy of the RIB is exchanged with loopback address as nexthop. And a static route can be configured to the loopback address to perform desired N-way ECMP path. M loopbacks are configured in this model, to achieve M different load balancing schemes: ECMP, weighted ECMP, Fast-reroute enabled paths etc.</t>
  <t>1 multi-hop EBGP session between loopback addresses, reachable via static route over the N links. Interface addresses are used as next-hops, without using additional loopbacks. 1 copy of the RIB is exchanged with MNH attribute to form N-way ECMP paths, weighted ECMP, Fast-reroute backup paths etc. BFD may be used to these directly connected BGP nexthops to detect liveness.</t>
</list></t>

</section>
<section anchor="flowspec-routes-with-multiple-redirect-ip-next-hops"><name>Flowspec Routes with Multiple "Redirect IP" next hops</name>

<t>There are existing protocol machinery which can benefit from the ability of MNH to clearly specify fallback behavior when multiple nexthops are involved. One example is the scenario described in <xref target="I-D.draft-ietf-idr-flowspec-redirect-ip"/> Sec 3 where multiple Redirect-to-IP nexthop addresses exist for a Flowspec prefix. In such a scenario, the receiving speakers may redirect the traffic to different nexthops, based on variables like IGP-cost. If instead, the MNH was used to specify the redirect-to-IP nexthop, then the order of preference between the different nexthops can be clearly specified using one flowspec route carrying a MNH containing those different nexthop-addresses specifying the desired preference-order. Such that, irrespective of IGP-cost, the receiving speakers will redirect the flow towards the same traffic collector device.</t>

</section>
<section anchor="color-only-resolution-next-hop"><name>Color-Only Resolution next hop</name>

<t>Another existing protocol machinery that manufactures nexthop addresses from overloaded extended color community is specified in <xref target="RFC9256"/> Sec 8.8.1. In a way, the color field is overloaded to carry one anycast BGP next-hop with pre-specified fallback options. This approach gives us only two next-hops to play with. The 'BGP nexthop address' and the 'Color-only nexthop'</t>

<t>Instead, the MNH could be used to achieve the same result with more flexibility. Multiple BGP nexthops can be carried, each resolving over a desired Transport class (Color), and with customizable fallback order. And the solution will work for non-SRTE networks as-well.</t>

</section>
<section anchor="problems-with-multihomed-pes-protecting-each-other"><name>Problems with Multihomed PEs Protecting Each Other</name>

<figure title="Example Topology with Multihomed PEs Protecting Each Other" anchor="topo-mh"><artwork><![CDATA[

                        +-----[PE11]    [RR1]
                        |       |        |
                        |       +------+ |
                     [CE1]             [P1]---[PE2]-----[CE2]
                        |       +------+
                        |       |
                        +-----[PE12]

                203.0.113.11                             203.0.113.22
                          ----  Traffic direction ---->

]]></artwork></figure>

<t>In a MPLS network, a router CE1 may be multihomed to two PEs PE11 and PE12. The PEs may re-advertise routes received from CE1 to the IBGP core with self as nexthop and a MPLS Label. The PEs may also protect failure of primary path to router CE1 by using the IBGP path via the other multihomed PE as a backup path. The advertised label has forwarding state installed with both primary and backup paths</t>

<t>Following problems are possible in this scenario:</t>

<section anchor="label-oscillation-between-multihomed-pes"><name>Label oscillation between Multihomed PEs</name>

<t>If "per nexthop" label allocation mechanism is used at the PEs, label allocation oscillation may occur when PE11 advertises a new label to PE12. Reception of a new label results in change of nexthop at PE12, as the received label is used as backup/repair nexthop leg, and per-nexthop label allocation is in use. Thus a new label is allocated by PE12 and advertised. And when this new label is received by the PE11, it allocates a new label in turn. This process repeats.</t>

<t>This problem can happen for either SAFI 4 or SAFI 128 routes.</t>

<t>This oscillation can be stopped only if the primary path label allocated by a PE does not depend on the primary path label advertised by other PE. A PE needs to be able to advertise multiple labels, one for use as primary path and another to be used as repair path by the receiver.</t>

<t>MNH attribute allows to advertise a Repair forwarding path label using <xref target="repair-path-tlv"/> in addition to Primary forwarding path label using <xref target="primary-path-tlv"/>. This avoids this label oscillation problem.</t>

</section>
<section anchor="forwarding-loop-between-multihomed-pes"><name>Forwarding loop between Multihomed PEs</name>

<t>If "per VRF table" label allocation mechanism is used at the PEs, a temporary forwarding loop may between PE11, PE12 in events like the CE1 router going down, which will cause both PE11-CE1 and PE12-CE1 links go down.</t>

<t>PE11 will forward traffic coming from PE2 on the backup path towards PE12. That packet will perform IP lookup in the VRF at PE12, which will result in the packet getting forwarded over the backup/repair path towards PE11. This loop will persist until global convergence completes, with the PEs send BGP withdrawals for the routes received from CE1 to each other.</t>

<t>This problem can happen for SAFI 128 routes or SAFI 1 routes, that perform IP lookup at egress PE..</t>

<t>This loop can also be avoided if the a PE can advertise a 'Repair path label' that does not include the primary path label advertised by other PE. A PE needs to be able to advertise multiple labels, one for use as primary path and another to be used as repair path by the receiver.</t>

<t>MNH attribute allows to advertise a Repair forwarding path label using <xref target="repair-path-tlv"/> in addition to Primary forwarding path label using <xref target="primary-path-tlv"/>. This avoids this forwarding loop problem also.</t>

</section>
</section>
<section anchor="signaling-intent-over-pe-ce-attachment-circuit"><name>Signaling Intent over PE-CE Attachment Circuit</name>

<t>BGP CT specifies procedures for Intent Driven Service Mapping in a service provider network, and defines 'Transport Class' construct to represent an Intent.</t>

<t>It may be desirable to allow a CE device to indicate in the data packet it sends what treatment it desires (the Intent) when the packet is forwarded within the provider network.</t>

<t>This section describes the mechanisms that enable such signaling. These procedures use existing AFIs 1 or 2, and service families (SAFI 1) on the PE-CE attachment circuit, with a new BGP attribute.</t>

<figure title="Example Topology with PE-CE Links" anchor="topo-pe-ce-intent"><artwork><![CDATA[
                                    ---Gold----->
                      [CE1]-----[PE1]---[P]----[PE2]-----[CE2]
                                    ---Bronze--->
                203.0.113.11                             203.0.113.22
                          ----  Traffic direction ---->

]]></artwork></figure>

<section anchor="using-dscp-in-multinexthop-attribute"><name>Using DSCP in MultiNexthop Attribute</name>

<t>Such an indication can be in form of DSCP code point (<xref target="RFC2474"></xref>) in the IP header.</t>

<t>In RFC2474, a Forwarding Class Selector maps to a PHB (Per-hop Behavior). The Transport Class construct is a PHB at transport layer.</t>

<t>Let PE1 be configured to map DSCP1 to Gold Transport class, and DSCP2 to Bronze Transport class. Based on the DSCP code point received on the IP traffic from CE1, PE1 forwards the IP packet over a Gold or Bronze tunnel. Thus, the forwarding is not based on just the destination IP address, but also the DSCP code point. This is known as Class Based Forwarding (CBF). Today CBF is configured at the PE1 device roles and CE1 doesn't receive any indication in BGP signaling regarding what DSCP code points are being offered by the provider network.</t>

<t>With a BGP MultiNexthop Attribute attached to a AFI/SAFI 1/1 service route, it is possible to extend the PE-CE BGP signaling (if used) to communicate such information to the CE1. In the preceding example, the MNH contains two Next hop Legs, described by two Forwarding Instruction TLVs. Each Next hop Leg contains PE1's peering self address in Endpoint Identifier TLV (<xref target="ep-tlv"/>), the color Gold or Bronze encoded in the Transport class ID TLV (<xref target="tc-tlv"/>), and associated DSCP code point indicating Gold or Bronze transport class encoded in the Payload Encapsulation Info TLV (<xref target="encap-tlv"/>). This allows the CE to discover what transport classes exist in the provider network, and which DSCP codepoint to encode so that traffic is forwarded using the desired transport class in the provided network.</t>

</section>
<section anchor="mpls-enabled-ce"><name>MPLS-enabled CE</name>

<t>If the PE-CE link is MPLS enabled, a distinct MPLS label can also be used to express Intent in data packets from CE. Enabling MPLS forwarding on PE-CE links comes with some security implications. This section gives details on these aspects.</t>

<t>Consider the ingress PE1 receiving a VPN prefix RD:Pfx1 received with VPN label VL1, next hop as PE2 and a mapping community containing TC1 as 'Transport class ID'. PE1 can allocate a MPLS Label PVL1 for the tuple "VPN Label, PNH Address, Transport class ID" and advertise to CE1.</t>

<t>Label PVL1 may identifies a service function at any node in the network, e.g. a Firewall device or egress node PE2. And, for the same service prefix, a distinct label may be advertised to different CEs, such that incoming traffic from different CEs to the same service prefix can be diverted to a distinct devices in the network for further processing. This provides Ingress Peer Engineering control to the network.</t>

<t>PE1 installs a MPLS FIB route for PVL1 with next hop as "Swap VL1, Push TL1 towards PE2". TL1 is the BGP CT label received for the tuple 'PE2, TC1'. In forwarding, when MPLS packet with label PVL1 is received from CE1, PVL1 Swaps to label VL1 and pushes the BGP CT label TL1. PE1 advertises the label "PVL1" in the MNH to CE1. PE1 forwards based on MPLS label without performing any IP lookup. This allows for PE1 to be a low IP FIB device and still support CBF by using MPLS Label inferred PHB. The number of MPLS Labels consumed at PE1 for this approach will be proportional to the number of Service functions and Intents that are exposed to CE1.</t>

<t>A BGP MultiNexthop Attribute is attached to a AFI/SAFI 1/1 service route to convey the MPLS Label information to CE1. In the preceding example, the MNH contains two Next hop Legs, described by two Forwarding Instruction TLVs. Each Next hop Leg contains PE1's peering self address in Endpoint Identifier TLV (<xref target="ep-tlv"/>), the color Gold or Bronze encoded in the Transport class ID TLV (<xref target="tc-tlv"/>), and associated MPLS Label "PVL1" or "PVL2" encoded in the Payload Encapsulation Info TLV (<xref target="encap-tlv"/>). This allows the CE to discover what transport classes exist in the provider network, and which MPLS Label to encode so that traffic is forwarded using the desired transport class.</t>

<section anchor="secure-mpls-forwarding-on-inter-as-link"><name>Secure MPLS Forwarding on Inter-AS Link</name>

<t>The MPLS enabled PE-CE attachment circuit is considered connecting to an untrusted domain. Such interfaces can be secured against MPLS label spoofing by a walled garden approach using "MPLS context tables".</t>

<t>The PE1-CE1 interface can be confined to a specific MPLS context table "A" corresponding to the BGP peer. Such that only the routes for labels advertised to CE1 are installed in MPLS context table "A".</t>

<t>This ensures that if CE1 sends MPLS packet with a label that was not advertised to the CE1, the packet will be dropped.</t>

<t>Furthermore, the routes for labels PVL1, PVL2 installed in MPLS context table "A" can match on 'Bottom of stack' bit being 'one', ensuring a MPLS packet is accepted from CE1 only if it has no more than one label in the label stack.</t>

<t>However, the PE itself may not be able to perform any checks based on inner payload in the MPLS packet since it performs label swap forwarding. Such inner payload based checks may be offloaded to a downstream node that forwards and processes inner payload, e.g. a IP FIB router. These security aspects should be considered when using MPLS enabled CE devices.</t>

</section>
</section>
</section>
<section anchor="pe-signal-mpls-label-for-ipv4-unicast-routes"><name>4PE - Signal MPLS Label for IPv4 Unicast routes</name>

<t>This section describes how MNH can be used to signal MPLS explicit null label in AFI/SAFI: 1/1 routes in a pure IPv6 core environment, to achieve 4PE.</t>

<figure title="4PE Network with Pure IPv6 Core" anchor="fig-topo-4pe"><artwork><![CDATA[
                                   [RR1]
                                     .
                                     |
                                     __
                                  _.(  )..
       [CE1] - [PE1]-[P1]    ..  (_  v6  _) .. [PE2] - [CE2]
                                   (._..)


                    <---- Traffic Direction ----
              P1: PHP node.
              PE1: Egress PE.
]]></artwork></figure>

<t><xref target="fig-topo-4pe"/> shows a 4PE network with pure IPv6 core, PE1 is the egress PE connected to penultimate hop node P1. PE1 to PE2 have some IPv6 core tunneling protocol like LDPv6. When PE1 has advertised Implicit Null label in LDPv6, some implementations of P1 may not be able to forward the inner IPv4 payload to PE1.</t>

<t>To solve this problem, PE1 needs to signal IPv4 Explicit NULL Label (Special Label 0) to PE2. PE2 will push this IPv4 Explicit NULL Label received in the MNH on the AFI/SAFI:1/1 route. Such that P1 does a MPLS Label swap operation and does not need to look into inner payload.</t>

<t>MNH can be used by PE1 on a AFI/SAFI: 1/1 route, to advertise the IPv4 Explicit Null label for the IPv4 Unicast service route. MPLS Label is encoded in the Payload Encapsulation Info TLV (<xref target="encap-tlv"/>).</t>

<t>This allows the network to provide clear separation of service and transport routes, and not overloading AFI/SAFI: 1/4 to carry the IPv4 service routes. Not mixing service and transport routes improves security and manageability aspects of the network.</t>

<t>An egress PE may not need to advertise IPv4 Explicit Null label for the IPv4 service route, if it does UHP label in LDPv6. This model using MNH provides a homogenous service layer (AFI/SAFI: 1/1) that accommodates differences in requirement of different PE and P routers. Only the PEs which are connected to P nodes that cannot handle the PHP situation need to advertise Label using MNH. The service layer is kept consistent in the network, and can seamlessly extend to multiple domains without needing redistribution between AFI/SAFIs.</t>

<t>Not mixing service and transport routes improves security and manageability aspects of the network.</t>


</section>
</section>

    <section anchor="contributors" numbered="false" toc="include" removeInRFC="false">
        <name>Contributors</name>
    <contact initials="M." surname="Jeyananth" fullname="Minto Jeyananth">
      <organization>HPE</organization>
      <address>
        <postal>
          <street>1133 Innovation Way</street>
          <city>Sunnyvale</city>
          <region>CA</region>
          <code>94089</code>
          <country>USA</country>
        </postal>
        <email>minto@hpe.com</email>
      </address>
    </contact>
    <contact initials="D." surname="Reshma" fullname="Reshma Das">
      <organization>HPE</organization>
      <address>
        <postal>
          <street>1133 Innovation Way</street>
          <city>Sunnyvale</city>
          <region>CA</region>
          <code>94089</code>
          <country>USA</country>
        </postal>
        <email>reshma.das@hpe.com</email>
      </address>
    </contact>
    <contact initials="V." surname="Natrajan" fullname="Natrajan Venkataraman">
      <organization>HPE</organization>
      <address>
        <postal>
          <street>1133 Innovation Way</street>
          <city>Sunnyvale</city>
          <region>CA</region>
          <code>94089</code>
          <country>USA</country>
        </postal>
        <email>natrajan.venkataraman@hpe.com</email>
      </address>
    </contact>
    </section>

  </back>

<!-- ##markdown-source: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-->

</rfc>

