RFC 8296 PROPOSED STANDARD MULTICAST · MPLS · ENCAPSULATION

BIER Encapsulation: Encapsulation for Bit Index Explicit Replication (BIER) in MPLS and Non-MPLS Networks

The exact header that lets a multicast packet carry a bit for every router that should receive it.

On rfc-editor.org ↗
PUBLISHED
January 2018
AUTHORS
IJ. Wijnands, E. Rosen, A. Dolganow, J. Tantsura, S. Aldrin, I. Meilik
STREAM
IETF
DEFAULT BITSTRING LENGTH
256 bits
BFR-ID RANGE
1 to 65535
ENTROPY FIELD
20 bits
NEXT PROTOCOL FIELD
6 bits
NON-MPLS ETHERTYPE
0xAB37
HEADER VERSION
0
DOI
10.17487/RFC8296
ERRATA
Reported errata

RFC text

Contents
Original text · RFC Editor View on rfc-editor.org ↗
Internet Engineering Task Force (IETF)                 IJ. Wijnands, Ed.
Request for Comments: 8296                           Cisco Systems, Inc.
Category: Experimental                                     E. Rosen, Ed.
ISSN: 2070-1721                                   Juniper Networks, Inc.
                                                             A. Dolganow
                                                                   Nokia
                                                             J. Tantsura
                                                              Individual
                                                               S. Aldrin
                                                            Google, Inc.
                                                               I. Meilik
                                                                Broadcom
                                                            January 2018


        Encapsulation for Bit Index Explicit Replication (BIER)
                     in MPLS and Non-MPLS Networks

Abstract

   Bit Index Explicit Replication (BIER) is an architecture that
   provides optimal multicast forwarding through a "multicast domain",
   without requiring intermediate routers to maintain any per-flow state
   or to engage in an explicit tree-building protocol.  When a multicast
   data packet enters the domain, the ingress router determines the set
   of egress routers to which the packet needs to be sent.  The ingress
   router then encapsulates the packet in a BIER header.  The BIER
   header contains a bit string in which each bit represents exactly one
   egress router in the domain; to forward the packet to a given set of
   egress routers, the bits corresponding to those routers are set in
   the BIER header.  The details of the encapsulation depend on the type
   of network used to realize the multicast domain.  This document
   specifies a BIER encapsulation that can be used in an MPLS network
   or, with slight differences, in a non-MPLS network.

Status of This Memo

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

   This document defines an Experimental Protocol for the Internet
   community.  This document is a product of the Internet Engineering
   Task Force (IETF).  It represents the consensus of the IETF
   community.  It has received public review and has been approved for
   publication by the Internet Engineering Steering Group (IESG).  Not
   all documents approved by the IESG are a candidate for any level of
   Internet Standard; see Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   https://www.rfc-editor.org/info/rfc8296.

Copyright Notice

   Copyright (c) 2018 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

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

Table of Contents

   1. Introduction ....................................................3
   2. BIER Header .....................................................5
      2.1. In MPLS Networks ...........................................5
           2.1.1. Encapsulation Initial Four Octets ...................5
                  2.1.1.1. The BIER-MPLS Label ........................5
                  2.1.1.2. Other Fields of the Initial Four Octets ....8
           2.1.2. Remainder of Encapsulation ..........................9
           2.1.3. Further Encapsulating a BIER Packet ................12
      2.2. In Non-MPLS Networks ......................................13
           2.2.1. Encapsulation Initial Four Octets ..................13
                  2.2.1.1. The BIFT-id ...............................13
                  2.2.1.2. Other Fields of the Initial Four Octets ...13
           2.2.2. Remainder of Encapsulation .........................14
           2.2.3. Further Encapsulating a BIER Packet ................15
   3. Imposing and Processing the BIER Encapsulation .................16
   4. IANA Considerations ............................................18
   5. IEEE Considerations ............................................18
   6. Security Considerations ........................................19
   7. References .....................................................20
      7.1. Normative References ......................................20
      7.2. Informative References ....................................21
   Acknowledgements ..................................................22
   Contributors ......................................................22
   Authors' Addresses ................................................24

1.  Introduction

   [RFC8279] describes a new architecture for the forwarding of
   multicast data packets.  Known as "Bit Index Explicit Replication"
   (BIER), that architecture provides optimal forwarding of multicast
   data packets through a "multicast domain".  It does so without
   requiring any explicit tree-building protocol and without requiring
   intermediate nodes to maintain any per-flow state.

   This document will use terminology defined in [RFC8279].

   A router that supports BIER is known as a "Bit-Forwarding Router"
   (BFR).  A "BIER domain" is a connected set of BFRs, each of which has
   been assigned a BFR-prefix.  A BFR-prefix is a routable IP address of
   a BFR and is used by BIER to identify a BFR.  A packet enters a BIER
   domain at a Bit-Forwarding Ingress Router (BFIR) and leaves the BIER
   domain at one or more Bit-Forwarding Egress Routers (BFERs).  As
   specified in [RFC8279], each BFR of a given BIER domain is
   provisioned to be in one or more "sub-domains" (SDs).  In the context
   of a given SD, each BFIR and BFER must have a BFR-id that is unique
   within that SD.  A BFR-id is just a number in the range [1,65535]
   that, relative to a BIER SD, identifies a BFR uniquely.

   As described in [RFC8279], BIER requires that multicast data packets
   be encapsulated with a header that provides the information needed to
   support the BIER forwarding procedures.  This information includes
   the SD to which the packet has been assigned, a Set Identifier (SI),
   a BitString, and a BitStringLength (BSL).  Together, these values are
   used to identify the set of BFERs to which the packet must be
   delivered.

   This document defines an encapsulation that can be used in either
   MPLS networks or non-MPLS networks.  However, the construction and
   processing of the BIER header are slightly different in MPLS networks
   than in non-MPLS networks.  In particular:

   o  The handling of certain fields in the encapsulation header (the
      "BIER header") is different, depending upon whether the underlying
      network is an MPLS network or not.

   o  In an MPLS network, the first four octets of a BIER header are
      also the bottom entry (the last four octets) of an MPLS label
      stack.

   The MPLS-based encapsulation is explained in detail in Section 2.1.
   The differences between the MPLS-based encapsulation and the non-MPLS
   encapsulation are explained in Section 2.2.

   Following the BIER header is the "payload".  The payload may be an
   IPv4 packet, an IPv6 packet, an Ethernet frame, an MPLS packet, or an
   Operations, Administration, and Maintenance (OAM) packet.  (The use
   of BIER with other payload types is also possible but is not further
   discussed in this document.)  The BIER header contains information
   (the Next Protocol field) identifying the type of the payload.

   If the payload is an MPLS packet, then an MPLS label stack
   immediately follows the BIER header.  The top label of this MPLS
   label stack may be either a downstream-assigned label [RFC3031] or an
   upstream-assigned label [RFC5331].

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
   "OPTIONAL" in this document are to be interpreted as described in
   BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

2.  BIER Header

   The BIER header is shown in Figure 1.

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |              BIFT-id                  | TC  |S|     TTL       |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |Nibble |  Ver  |  BSL  |              Entropy                  |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |OAM|Rsv|    DSCP   |   Proto   |            BFIR-id            |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                BitString  (first 32 bits)                     ~
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     ~                                                               ~
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     ~                BitString  (last 32 bits)                      |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                           Figure 1: BIER Header

   The BIFT-id represents a particular Bit Index Forwarding
   Table (BIFT); see Section 6.4 of [RFC8279].  As explained in
   [RFC8279], each BIFT corresponds to a particular combination of SD,
   BSL, and SI.

   Section 2.1 explains how the fields of the encapsulation header are
   used in MPLS networks.  For those fields that are used differently in
   non-MPLS networks, Section 2.2 explains the differences.

   The default BitStringLength value for the encapsulations defined in
   this document is 256.  See Section 3 of [RFC8279] for a discussion of
   the default BitStringLength value.

2.1.  In MPLS Networks

2.1.1.  Encapsulation Initial Four Octets

2.1.1.1.  The BIER-MPLS Label

   As stated in [RFC8279], when a BIER domain is also an IGP domain, IGP
   extensions can be used by each BFR to advertise the BFR-id and
   BFR-prefix.  The extensions for OSPF are given in
   [OSPF_BIER_EXTENSIONS].  The extensions for IS-IS are given in
   [ISIS_BIER_EXTENSIONS].

   When a particular BIER domain is both an IGP domain and an MPLS
   network, we assume that each BFR will also use IGP extensions to
   advertise a set of one or more "BIER-MPLS" labels.  When the domain
   contains a single SD, a given BFR needs to advertise one such label
   for each combination of SI and BSL.  If the domain contains multiple
   SDs, a BFR needs to advertise one such label per SI per BSL for
   each SD.

   In some environments, the only routing protocol in a BIER domain
   might be BGP; in this case, the BGP extensions described in
   [BGP_BIER_EXTENSIONS] can be used to advertise the necessary set of
   BIER-MPLS labels.

   The BIER-MPLS labels are locally significant (i.e., unique only to
   the BFR that advertises them) downstream-assigned MPLS labels.
   Penultimate hop popping [RFC3031] MUST NOT be applied to a BIER-MPLS
   label.

   Suppose, for example, that there is a single SD (the default SD),
   that the network is using a BSL of 256, and that all BFERs in the SD
   have BFR-ids in the range [1,512].  Since each BIER BitString is 256
   bits long, this requires the use of two SIs: SI=0 and SI=1.  So each
   BFR will advertise, via IGP extensions, two MPLS labels for BIER: one
   corresponding to SI=0 and one corresponding to SI=1.  The
   advertisements of these labels will also bind each label to the
   default SD and to BSL 256.

   As another example, suppose a particular BIER domain contains two SDs
   (SD 0 and SD 1), supports two BSLs (256 and 512), and contains
   1024 BFRs.  A BFR that is provisioned for both SDs, and that supports
   both BSLs, would have to advertise the following set of BIER-MPLS
   labels:

      L1:   corresponding to SD 0, BSL 256, SI 0.

      L2:   corresponding to SD 0, BSL 256, SI 1.

      L3:   corresponding to SD 0, BSL 256, SI 2.

      L4:   corresponding to SD 0, BSL 256, SI 3.

      L5:   corresponding to SD 0, BSL 512, SI 0.

      L6:   corresponding to SD 0, BSL 512, SI 1.

      L7:   corresponding to SD 1, BSL 256, SI 0.

      L8:   corresponding to SD 1, BSL 256, SI 1.

      L9:   corresponding to SD 1, BSL 256, SI 2.

      L10:  corresponding to SD 1, BSL 256, SI 3.

      L11:  corresponding to SD 1, BSL 512, SI 0.

      L12:  corresponding to SD 1, BSL 512, SI 1.

   The above example should not be taken as implying that the BFRs need
   to advertise 12 individual labels.  For instance, instead of
   advertising a label for <SD 1, BSL 512, SI 0> and a label for
   <SD 1, BSL 512, SI 1>, a BFR could advertise a contiguous range of
   labels (in this case, a range containing exactly two labels)
   corresponding to <SD 1, BSL 512>.  The first label in the range could
   correspond to SI 0, and the second to SI 1.  The precise mechanism
   for generating and forming the advertisements is outside the scope of
   this document; see [OSPF_BIER_EXTENSIONS] and [ISIS_BIER_EXTENSIONS].

   The BIER-MPLS label corresponding to a particular combination of SD,
   SI, and BSL is interpreted as representing the BIFT that corresponds
   to that same combination of SD, SI, and BSL.  That is, the BIER-MPLS
   label performs the function of a BIFT-id.  This label value is
   carried in the BIFT-id field of the BIER encapsulation.

   It is crucial to understand that in an MPLS network the first
   four octets of the BIER encapsulation header are also the last
   four octets of the MPLS header.  Therefore, any prior MPLS label
   stack entries MUST have the S bit (see [RFC3032]) clear (i.e., the
   S bit must be 0).

   When a BFR receives an MPLS packet and the next label to be processed
   is one of its BIER-MPLS labels, it will assume that the remainder of
   the BIER header (see Section 2.1.2) immediately follows the stack.

   Note that in practice, labels only have to be assigned if they are
   going to be used.  If a particular BIER domain supports BSLs 256 and
   512, but some SD, say SD 1, only uses BSL 256, then it is not
   necessary to assign labels that correspond to the combination of SD 1
   and BSL 512.

2.1.1.2.  Other Fields of the Initial Four Octets

   TC:

      The "Traffic Class" field [RFC5462] has its usual meaning in an
      MPLS label stack entry.

   S bit:

      When a BIER packet is traveling through an MPLS network, the
      high-order 20 bits of the initial four octets of the BIER
      encapsulation contain an MPLS label in the BIFT-id field.  These
      four octets are treated as the final entry in the packet's MPLS
      label stack.  Hence, the S bit (see [RFC3032]) MUST be set to 1.
      If there are any MPLS label stack entries immediately preceding
      the BIER encapsulation, the S bit of those label stack entries
      MUST be set to 0.

   TTL:

      This is the usual MPLS "Time to Live" field [RFC3032].  When a
      BIER packet is received, its "incoming TTL" (see below) is taken
      from this TTL field.

      When a BIER packet is forwarded to one or more BFR adjacencies,
      the BIER-MPLS label carried by the forwarded packet MUST have a
      TTL field whose value is one less than that of the packet's
      incoming TTL.

      If a BIER packet's incoming TTL is 1 or greater and one of the
      bits in its BitString identifies the current BFR, then the current
      BFR is a BFER for the packet.  Therefore, the current BFR MUST
      process the packet as a BFER, e.g., by removing the BIER
      encapsulation and processing the payload based on the contents of
      the Proto (Next Protocol) field.

      If the incoming TTL is 0, the packet is considered to be
      "expired".  If the incoming TTL is 1 and the BitString has a bit
      set that does not identify the current BFR, the packet is also
      considered to be expired.  Expired packets SHOULD be passed to an
      error-handling procedure.  (Optional implementation-specific
      rate limiting may be applied to control the rate at which packets
      are passed to the error-handling procedure.)  Specification of the
      error-handling procedure is outside the scope of this document.

      Note that if a received BIER packet has an incoming TTL of 1 and
      its BitString has a bit set identifying the current BFR, the
      payload MUST be processed by the current BFR, but the packet
      MUST NOT be forwarded further, and the packet SHOULD also be
      passed to the error-handling procedures for expired packets
      (subject to any implementation-specific rate limiting).

2.1.2.  Remainder of Encapsulation

   Nibble:

      This field is set to the binary value 0101; this ensures that the
      MPLS ECMP logic will not confuse the remainder of the BIER header
      with an IP header or with the header of a pseudowire packet.  In
      an MPLS network, if a BFR receives a BIER packet with any other
      value in the first nibble after the label stack, it SHOULD discard
      the packet and log an error.

   Ver:

      This 4-bit field identifies the version of the BIER header.  This
      document specifies version 0 of the BIER header.  If a packet is
      received by a particular BFR and that BFR does not support the
      specified version of the BIER header, the BFR MUST discard the
      packet and log an error.

      The value 0xF is reserved for experimental use; that value
      MUST NOT be assigned by any future IETF document or by IANA.

   BSL:

      This 4-bit field encodes the length in bits of the BitString.

      Note: When parsing the BIER header, a BFR MUST infer the length of
      the BitString from the BIFT-id and MUST NOT infer it from the
      value of this field.  This field is present only to enable offline
      tools (such as LAN analyzers) to parse the BIER header.

      If k is the length of the BitString, the value of this field is
      log2(k)-5.  However, only certain values are supported:

         1: 64 bits

         2: 128 bits

         3: 256 bits

         4: 512 bits

         5: 1024 bits

         6: 2048 bits

         7: 4096 bits

      The value of this field MUST NOT be set to any value other than
      those listed above.  A received packet containing another value in
      this field SHOULD be discarded and an error logged.  If the value
      in this field is other than what is expected based on the
      BIER-MPLS label, the packet SHOULD be discarded and an error
      logged.

   Entropy:

      This 20-bit field specifies an "entropy" value that can be used
      for load-balancing purposes.  The BIER forwarding process may do
      equal-cost load balancing, in which case the load-balancing
      procedure MUST choose the same path for any two packets that have
      the same entropy value and the same BitString.  Please see
      Section 6.7 ("Equal-Cost Multipath Forwarding") of [RFC8279] for a
      more detailed discussion of BIER load-balancing procedures.

      If a BFIR is encapsulating (as the payload) MPLS packets that have
      entropy labels, the BFIR MUST ensure that if two such packets have
      the same MPLS entropy label they also have the same value of the
      BIER entropy field.

   OAM:

      By default, these two bits are set to 0 by the BFIR and are not
      modified by other BFRs.  These two bits have no effect on the path
      taken by a BIER packet and have no effect on the quality of
      service applied to a BIER packet.

      The use of these bits in other than the default manner is
      OPTIONAL.  Specification of the non-default use or uses of these
      bits is outside the scope of this document; see [BIER-PMM] for an
      example of such a specification.

   Rsv:

      These two bits are currently unused.  They SHOULD be set to 0 upon
      transmission and MUST be ignored upon reception.

   DSCP:

      By default, this 6-bit field is not used in MPLS networks.  The
      default behavior is that all six bits SHOULD be set to 0 upon
      transmission and MUST be ignored upon reception.

      Non-default use of this field in MPLS networks is outside the
      scope of this document.

   Proto:

      This 6-bit "Next Protocol" field identifies the type of the
      payload.  (The "payload" is the packet or frame immediately
      following the BIER header.)  IANA has created a registry called
      "BIER Next Protocol Identifiers".  This field is to be populated
      with the appropriate entry from that registry.

      If a BFER receives a BIER packet but does not recognize (or does
      not support) the value of the Next Protocol field, the BFER SHOULD
      discard the packet and log an error.

   BFIR-id:

      By default, this is the BFR-id of the BFIR, in the SD to which the
      packet has been assigned.  The BFR-id is encoded in the 16-bit
      field as an unsigned integer in the range [1,65535].

      Certain applications may require that the BFIR-id field contain
      the BFR-id of a BFR other than the BFIR.  However, that usage of
      the BFIR-id field is outside the scope of this document.

   BitString:

      This field holds the BitString that, together with the packet's SI
      and SD, identifies the destination BFERs for this packet.  Note
      that the SI and SD for the packet are not carried explicitly in
      the BIER header, as a particular BIFT-id always corresponds to a
      particular SI and SD.

2.1.3.  Further Encapsulating a BIER Packet

   Sending a BIER packet from one BFR to another may require the packet
   to be further encapsulated.  For example, in some scenarios it may be
   necessary to encapsulate a BIER packet in an Ethernet frame; in other
   scenarios it may be necessary to encapsulate a BIER packet in a UDP
   packet.  In such cases, the BIER packet itself is the payload of an
   "outer" encapsulation.

   In this document, we assume that the frame or packet carrying a BIER
   packet as its payload is a unicast frame or packet.  That is,
   although a BIER packet is a multicast packet, we assume that the
   frame or packet carrying the BIER packet as its payload is unicast
   from one BFR to the next.

   Generally, the outer encapsulation has a codepoint identifying the
   "next protocol".  The outer encapsulation's "next protocol" codepoint
   for MPLS MUST be used.  If a particular outer encapsulation has a
   codepoint for "MPLS with downstream-assigned label" and a different
   codepoint for "MPLS with upstream-assigned label", the codepoint for
   "MPLS with downstream-assigned label" MUST be used.

   For example, if a BIER packet is encapsulated in an Ethernet frame,
   the Ethertype MUST be 0x8847 [RFC5332], which is the Ethertype for a
   unicast Ethernet frame that carries an MPLS packet whose label stack
   begins with a downstream-assigned label.

   In the special case where the outer encapsulation is MPLS, the outer
   encapsulation has no "next protocol" codepoint.  All that is needed
   to encapsulate the BIER packet is to push more MPLS label stack
   entries (with the S bit clear) on the BIER packet's label stack.

   If two BIER packets have the same value in the entropy field of their
   respective BIER headers and if both are placed in an outer
   encapsulation, it is desirable for the outer encapsulation to
   preserve the fact that the two packets have the same entropy.  If the
   outer encapsulation is MPLS and if the MPLS entropy label [RFC6790]
   is in use in a given deployment, one way to do this is to copy the
   value of the BIER header entropy field into an MPLS entropy label.

2.2.  In Non-MPLS Networks

2.2.1.  Encapsulation Initial Four Octets

2.2.1.1.  The BIFT-id

   In non-MPLS networks, a BIFT-id MUST be assigned for every
   combination of <SD, SI, BSL> that is to be used in that network.  The
   correspondence between a BIFT-id and a particular <SD, SI, BSL>
   triple is unique throughout the BIER domain and is known to all the
   BFRs in the BIER domain.

   The means by which the BIFT-ids are assigned, and the means by which
   these assignments are made known to the BFRs, are outside the scope
   of this document.

   In an MPLS network, since the BIFT-id is an MPLS label, its value may
   be changed as a BIER packet goes from BFR to BFR.  In a non-MPLS
   network, since the BIFT-id is domain-wide unique, it is not expected
   to change as a BIER packet travels.

2.2.1.2.  Other Fields of the Initial Four Octets

   TC:

      By default, the TC field has no significance in a non-MPLS
      network.  The default behavior is that this field SHOULD be set to
      the binary value 000 upon transmission and MUST be ignored upon
      reception.

      Non-default use of this field in non-MPLS networks is outside the
      scope of this document.

   S bit:

      The S bit has no significance in a non-MPLS network.  It SHOULD be
      set to 1 upon transmission, but it MUST be ignored upon reception.

   TTL:

      This is the BIER "Time to Live" field.  Its purpose is to prevent
      BIER packets from looping indefinitely in the event of improper
      operation of the control plane.  When a BIER packet is received,
      its "incoming TTL" (see below) is taken from this TTL field.

      The effect of this field on the processing of a BIER packet is
      described in Section 2.1.1.2.

2.2.2.  Remainder of Encapsulation

   Nibble:

      This field SHOULD be set to 0000 upon transmission but MUST be
      ignored upon reception.

   Ver:

      See Section 2.1.2.

   BSL:

      See Section 2.1.2.

   Entropy:

      See Section 2.1.2.

   OAM:

      See Section 2.1.2.

   Rsv:

      See Section 2.1.2.

   DSCP:

      This 6-bit field MAY be used to hold a Differentiated Services
      Codepoint [RFC2474].  The significance of this field is outside
      the scope of this document.

   Proto:

      See Section 2.1.2.

   BFIR-id:

      See Section 2.1.2.

   BitString:

      See Section 2.1.2.

2.2.3.  Further Encapsulating a BIER Packet

   Sending a BIER packet from one BFR to another may require the packet
   to be further encapsulated.  For example, in some scenarios it may be
   necessary to encapsulate a BIER packet in an Ethernet frame; in other
   scenarios it may be necessary to encapsulate a BIER packet in a UDP
   packet.  In such cases, the BIER packet itself is the payload of an
   "outer" encapsulation.

   In this document, we assume that the frame or packet carrying a BIER
   packet as its payload is a unicast frame or packet.  That is,
   although a BIER packet is a multicast packet, we assume that the
   frame or packet carrying the BIER packet as its payload is unicast
   from one BFR to the next.

   Generally, the outer encapsulation has a codepoint identifying the
   "next protocol".  This codepoint MUST be set to a value that means
   "non-MPLS BIER".  In particular, a codepoint that means "MPLS" (with
   either upstream-assigned or downstream-assigned labels) MUST NOT
   be used.

   By requiring the use of a distinct codepoint for "non-MPLS BIER", we
   allow for deployment scenarios where non-MPLS BIER can coexist with
   non-BIER MPLS.  The BIFT-id values used by the former will not
   conflict with MPLS label values used by the latter.

   Therefore, if a non-MPLS BIER packet is encapsulated in an Ethernet
   header, the Ethertype MUST NOT be 0x8847 or 0x8848 [RFC5332].  IEEE
   has assigned Ethertype 0xAB37 for non-MPLS BIER packets.

   In the special case where the outer encapsulation is MPLS, the outer
   encapsulation has no "next protocol" codepoint.  If it is necessary
   to use MPLS as an outer encapsulation for BIER packets, it is
   RECOMMENDED to use the MPLS encapsulation for BIER.  Procedures for
   encapsulating a non-MPLS BIER packet in MPLS are outside the scope of
   this document.

   If two BIER packets have the same value in the entropy field of their
   respective BIER headers and if both are placed in an outer
   encapsulation, it is desirable for the outer encapsulation to
   preserve the fact that the two packets have the same entropy.

3.  Imposing and Processing the BIER Encapsulation

   Each BFIR is expected to know the Maximum Transmission Unit (MTU) of
   the BIER domain.  This may be known by provisioning, or by some other
   method outside the scope of this document.  Each BFIR also knows the
   size of the BIER encapsulation.  Thus, each BFIR can deduce the
   maximum size of the payload that can be encapsulated in a BIER
   packet.  We will refer to this payload size as the BIER-MTU.

   If a BFIR receives a multicast packet from outside the BIER domain
   and the packet size exceeds the BIER-MTU, the BFIR takes whatever
   action is appropriate to take when receiving a multicast packet that
   is too large to be forwarded to all its next hops.  If the
   appropriate action is to drop the packet and advertise an MTU to the
   source, then the BFIR drops the packet and advertises the BIER-MTU.
   If the appropriate action is to fragment the packet, then the
   procedures of this section are applied, in sequence, to each
   fragment.

   When a BFIR processes a multicast packet (or fragment thereof) from
   outside the BIER domain, the BFIR carries out the following
   procedure:

   1.  By consulting the "multicast flow overlay" [RFC8279], it
       determines the value of the Proto field.

   2.  By consulting the multicast flow overlay, it determines the set
       of BFERs that must receive the packet.

   3.  If more than one SD is supported, the BFIR assigns the packet to
       a particular SD.  Procedures for determining the SD to which a
       particular packet should be assigned are outside the scope of
       this document.

   4.  The BFIR looks up the BFR-id, in the given SD, of each of the
       BFERs.

   5.  The BFIR converts each such BFR-id into "SI:BitString" format, as
       described in [RFC8279].

   6.  All such BFR-ids that have the same SI can be encoded into the
       same BitString.  Details of this encoding can be found in
       [RFC8279].  For each distinct SI that occurs in the list of the
       packet's destination BFERs:

       a.  The BFIR makes a copy of the multicast data packet and
           encapsulates the copy in a BIER header (see Section 2).  The
           BIER header contains the BitString that represents all the
           destination BFERs whose BFR-ids (in the given SD) correspond
           to the given SI.  It also contains the BFIR's BFR-id in the
           SD to which the packet has been assigned.

           Note well that for certain applications it may be necessary
           for the BFIR-id field to contain the BFR-id of a BFR other
           than the BFIR that is creating the header.  Such uses are
           outside the scope of this document.

       b.  The BFIR then applies to that copy the forwarding procedure
           of [RFC8279].  This may result in one or more copies of the
           packet (possibly with a modified BitString) being transmitted
           to a neighboring BFR.

       c.  If the non-MPLS BIER encapsulation is being used, the BIFT-id
           field is set to the BIFT-id that corresponds to the packet's
           <SD, SI, BSL>.  The TTL is set according to policy.

           If the MPLS BIER encapsulation is being used, the BFIR finds
           the BIER-MPLS label that was advertised by the neighbor as
           corresponding to the given <SD, SI, BSL>.  An MPLS label
           stack is then prepended to the packet.  This label stack
           [RFC3032] will contain one label -- the aforementioned
           BIER-MPLS label.  The S bit MUST be set, indicating the end
           of the MPLS label stack.  The TTL field of this label stack
           entry is set according to policy.

       d.  The packet may then be transmitted to the neighboring BFR.
           (In an MPLS network, this may result in additional MPLS
           labels being pushed on the stack.  For example, if an RSVP-TE
           tunnel is used to transmit packets to the neighbor, a label
           representing that tunnel would be pushed onto the stack.)

   When an intermediate BFR is processing a received MPLS packet and one
   of the BFR's own BIER-MPLS labels rises to the top of the label
   stack, the BFR infers the BSL from the label.  The SI and SD are also
   implicitly identified by the label.  The BFR then follows the
   forwarding procedures of [RFC8279].  If it forwards a copy of the
   packet to a neighboring BFR, it first swaps the label at the top of
   the label stack with the BIER-MPLS label, advertised by that
   neighbor, that corresponds to the same <SD, SI, BSL>.  Note that when
   this swap operation is done, the TTL field of the BIER-MPLS label of
   the outgoing packet MUST be one less than the "incoming TTL" of the
   packet, as defined in Section 2.1.1.2.

   When an intermediate BFR is processing a received non-MPLS BIER
   packet, the BFR infers the BSL from the BIFT-id.  The SI and SD are
   also implicitly identified by the BIFT-id.  The BFR then follows the
   forwarding procedures of [RFC8279].

   If the BIER payload is an MPLS packet, the BIER header is followed by
   an MPLS label stack.  This stack is separate from any MPLS stack that
   may precede the BIER header.  For an example of an application where
   it is useful to carry an MPLS packet as the BIER payload, see
   [BIER_MVPN].  If the BIER encapsulation's Proto field indicates that
   the payload is an MPLS packet with an upstream-assigned label at the
   top of the stack, the upstream-assigned label is interpreted in the
   context of <BFIR-id, sub-domain-id>.  Note that the sub-domain-id
   must be inferred from the BIFT-id.

4.  IANA Considerations

   IANA has set up a registry called "BIER Next Protocol Identifiers".
   The registration policy for this registry is "IETF Review" [RFC8126]
   [RFC7120].

   The initial values in the "BIER Next Protocol Identifiers"
   registry are:

   0: Reserved

   1: MPLS packet with downstream-assigned label at top of stack

   2: MPLS packet with upstream-assigned label at top of stack

   3: Ethernet frame

   4: IPv4 packet

   5: OAM packet (Reference: [BIER_PING])

   6: IPv6 packet

   63:  Reserved

5.  IEEE Considerations

   IEEE has assigned Ethertype 0xAB37 for non-MPLS BIER packets.

6.  Security Considerations

   Insofar as this document makes use of MPLS, it inherits any security
   considerations that apply to the use of the MPLS data plane.

   If a BIER encapsulation header is modified in ways other than those
   specified in [RFC8279] and in this document, packets may be lost,
   stolen, or otherwise misdelivered.  Such modifications are likely to
   go undetected, as the BIER encapsulation does not provide
   cryptographic integrity protection.

   Layer 2 encryption can be used to ensure that a BIER-encapsulated
   packet is not altered while in transit between adjacent BFRs.  If a
   BFR itself is compromised, there is no way to prevent the compromised
   BFR from making illegitimate modifications to the BIER header or to
   prevent it from misforwarding or misdelivering the BIER-encapsulated
   packet.

   If the routing underlay (see Section 4.1 of [RFC8279]) is based on a
   unicast routing protocol, BIER assumes that the routers participating
   in the unicast routing protocol have not been compromised.  BIER has
   no procedures to ensure that the unicast routing adjacencies have not
   been compromised; that falls within the scope of whatever unicast
   routing protocols are being used.

   BIER-encapsulated packets should generally not be accepted from
   untrusted interfaces or tunnels.  For example, an operator may wish
   to have a policy of accepting BIER-encapsulated packets only from
   interfaces to trusted routers, and not from customer-facing
   interfaces.

   There may be applications that require a BFR to accept a
   BIER-encapsulated packet from an interface to a system that is not
   controlled by the network operator.  For instance, there may be an
   application in which a virtual machine in a data center submits
   BIER-encapsulated packets to a router.  In such a case, it is
   desirable to verify that the packet is from a legitimate source and
   that its BitString denotes only systems to which that source is
   allowed to send.  However, the BIER encapsulation itself does not
   provide a way to verify that the source is (1) legitimate, (2) really
   the system denoted by the BFIR-id, or (3) allowed to set any
   particular set of bits in the BitString.

   Insofar as this document relies upon IGP extensions, it inherits any
   security considerations that apply to the IGP.

   The security considerations of [RFC8279] also apply.

7.  References

7.1.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/info/rfc2119>.

   [RFC2474]  Nichols, K., Blake, S., Baker, F., and D. Black,
              "Definition of the Differentiated Services Field
              (DS Field) in the IPv4 and IPv6 Headers", RFC 2474,
              DOI 10.17487/RFC2474, December 1998,
              <https://www.rfc-editor.org/info/rfc2474>.

   [RFC3031]  Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol
              Label Switching Architecture", RFC 3031,
              DOI 10.17487/RFC3031, January 2001,
              <https://www.rfc-editor.org/info/rfc3031>.

   [RFC3032]  Rosen, E., Tappan, D., Fedorkow, G., Rekhter, Y.,
              Farinacci, D., Li, T., and A. Conta, "MPLS Label Stack
              Encoding", RFC 3032, DOI 10.17487/RFC3032, January 2001,
              <https://www.rfc-editor.org/info/rfc3032>.

   [RFC5331]  Aggarwal, R., Rekhter, Y., and E. Rosen, "MPLS Upstream
              Label Assignment and Context-Specific Label Space",
              RFC 5331, DOI 10.17487/RFC5331, August 2008,
              <https://www.rfc-editor.org/info/rfc5331>.

   [RFC5332]  Eckert, T., Rosen, E., Ed., Aggarwal, R., and Y. Rekhter,
              "MPLS Multicast Encapsulations", RFC 5332,
              DOI 10.17487/RFC5332, August 2008,
              <https://www.rfc-editor.org/info/rfc5332>.

   [RFC5462]  Andersson, L. and R. Asati, "Multiprotocol Label Switching
              (MPLS) Label Stack Entry: "EXP" Field Renamed to "Traffic
              Class" Field", RFC 5462, DOI 10.17487/RFC5462,
              February 2009, <https://www.rfc-editor.org/info/rfc5462>.

   [RFC7120]  Cotton, M., "Early IANA Allocation of Standards Track Code
              Points", BCP 100, RFC 7120, DOI 10.17487/RFC7120,
              January 2014, <https://www.rfc-editor.org/info/rfc7120>.

   [RFC8126]  Cotton, M., Leiba, B., and T. Narten, "Guidelines for
              Writing an IANA Considerations Section in RFCs", BCP 26,
              RFC 8126, DOI 10.17487/RFC8126, June 2017,
              <https://www.rfc-editor.org/info/rfc8126>.

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in
              RFC 2119 Key Words", BCP 14, RFC 8174,
              DOI 10.17487/RFC8174, May 2017,
              <https://www.rfc-editor.org/info/rfc8174>.

   [RFC8279]  Wijnands, IJ., Ed., Rosen, E., Ed., Dolganow, A.,
              Przygienda, T., and S. Aldrin, "Multicast Using Bit Index
              Explicit Replication (BIER)", RFC 8279,
              DOI 10.17487/RFC8279, November 2017,
              <https://www.rfc-editor.org/info/rfc8279>.

7.2.  Informative References

   [BGP_BIER_EXTENSIONS]
              Xu, X., Ed., Chen, M., Patel, K., Wijnands, IJ., and A.
              Przygienda, "BGP Extensions for BIER", Work in Progress,
              draft-ietf-bier-idr-extensions-04, January 2018.

   [BIER-PMM] Mirsky, G., Zheng, L., Chen, M., and G. Fioccola,
              "Performance Measurement (PM) with Marking Method in Bit
              Index Explicit Replication (BIER) Layer", Work in
              Progress, draft-ietf-bier-pmmm-oam-03, October 2017.

   [BIER_MVPN]
              Rosen, E., Ed., Sivakumar, M., Aldrin, S., Dolganow, A.,
              and T. Przygienda, "Multicast VPN Using BIER", Work in
              Progress, draft-ietf-bier-mvpn-09, November 2017.

   [BIER_PING]
              Kumar, N., Pignataro, C., Akiya, N., Zheng, L., Chen, M.,
              and G. Mirsky, "BIER Ping and Trace", Work in Progress,
              draft-ietf-bier-ping-02, July 2017.

   [ISIS_BIER_EXTENSIONS]
              Ginsberg, L., Ed., Przygienda, A., Aldrin, S., and J.
              Zhang, "BIER support via ISIS", Work in Progress,
              draft-ietf-bier-isis-extensions-06, October 2017.

   [OSPF_BIER_EXTENSIONS]
              Psenak, P., Ed., Kumar, N., Wijnands, IJ., Dolganow, A.,
              Przygienda, T., Zhang, J., and S. Aldrin, "OSPF Extensions
              for BIER", Work in Progress, draft-ietf-bier-ospf-bier-
              extensions-10, December 2017.

   [RFC6790]  Kompella, K., Drake, J., Amante, S., Henderickx, W., and
              L. Yong, "The Use of Entropy Labels in MPLS Forwarding",
              RFC 6790, DOI 10.17487/RFC6790, November 2012,
              <https://www.rfc-editor.org/info/rfc6790>.

Acknowledgements

   The authors wish to thank Rajiv Asati, John Bettink, Nagendra Kumar,
   Christian Martin, Neale Ranns, Greg Shepherd, Ramji Vaithianathan,
   Xiaohu Xu, and Jeffrey Zhang for their ideas and contributions to
   this work.

Contributors

   The following people (listed in alphabetical order) contributed
   significantly to the content of this document and should be
   considered co-authors:

   Mach(Guoyi) Chen
   Huawei
   Email: mach.chen@huawei.com

   Arkadiy Gulko
   Thomson Reuters
   195 Broadway
   New York, NY  10007
   United States of America
   Email: arkadiy.gulko@thomsonreuters.com

   Wim Henderickx
   Nokia
   Copernicuslaan 50
   Antwerp  2018
   Belgium
   Email: wim.henderickx@nokia.com

   Martin Horneffer
   Deutsche Telekom
   Hammer Str. 216-226
   Muenster  48153
   Germany
   Email: Martin.Horneffer@telekom.de
   Uwe Joorde
   Deutsche Telekom
   Hammer Str. 216-226
   Muenster  D-48153
   Germany
   Email: Uwe.Joorde@telekom.de

   Tony Przygienda
   Juniper Networks, Inc.
   1194 N. Mathilda Ave.
   Sunnyvale, California  94089
   United States of America
   Email: prz@juniper.net

Authors' Addresses

   IJsbrand Wijnands (editor)
   Cisco Systems, Inc.
   De Kleetlaan 6a
   Diegem  1831
   Belgium
   Email: ice@cisco.com

   Eric C. Rosen (editor)
   Juniper Networks, Inc.
   10 Technology Park Drive
   Westford, Massachusetts  01886
   United States of America
   Email: erosen@juniper.net

   Andrew Dolganow
   Nokia
   438B Alexandra Rd #08-07/10
   Alexandra Technopark
   Singapore  119968
   Singapore
   Email: andrew.dolganow@nokia.com

   Jeff Tantsura
   Individual
   Email: jefftant.ietf@gmail.com

   Sam K. Aldrin
   Google, Inc.
   1600 Amphitheatre Parkway
   Mountain View, California  94043
   United States of America
   Email: aldrin.ietf@gmail.com

   Israel Meilik
   Broadcom
   Email: israel@broadcom.com

ISO Plain Language

On this page
PROBLEMBIER forwards multicast without per-flow state, but routers need to agree on one header format that carries the set of egress routers, in both MPLS and non-MPLS networks.
MECHANISMThe ingress router wraps each packet in a BIER header whose BitString has one bit per egress router. In MPLS the first four octets double as the bottom MPLS label stack entry; in non-MPLS networks a domain-wide BIFT-id takes the label's place.
WHY IT MATTERSThe encapsulation lets every router read the BitString, find the right forwarding table and replicate the packet as needed. It has no cryptographic protection, so BIER packets should not be accepted from untrusted interfaces.

§1Introduction

The BIER architecture describes a new approach to the forwarding of multicast data packets. Known as 'Bit Index Explicit Replication', it provides optimal forwarding of multicast data packets through a multicast domain, without requiring any explicit tree-building protocol and without requiring intermediate nodes to maintain any per-flow state.

A router that supports BIER is known as a Bit-Forwarding Router. A BIER domain is a connected set of such routers, each of which has been assigned a BFR-prefix, a routable IP address used to identify it. A packet enters a BIER domain at a Bit-Forwarding Ingress Router and leaves at one or more Bit-Forwarding Egress Routers. Each router of a given BIER domain is provisioned to be in one or more sub-domains, and in the context of a given sub-domain each ingress and egress router must have a BFR-id that is unique within that sub-domain. A BFR-id is a number in the range [1,65535] that identifies a router uniquely within a sub-domain.

BIER requires that multicast data packets be encapsulated with a header that provides the information needed to support the BIER forwarding procedures. That information includes the sub-domain to which the packet has been assigned, a Set Identifier, a BitString, and a BitStringLength. Together these values identify the set of egress routers to which the packet must be delivered.

This document defines an encapsulation that can be used in either MPLS networks or non-MPLS networks, but the construction and processing of the BIER header are slightly different in the two. The handling of certain fields is different depending on whether the underlying network is an MPLS network, and in an MPLS network the first four octets of a BIER header are also the bottom entry of an MPLS label stack.

Following the BIER header is the payload, which may be an IPv4 packet, an IPv6 packet, an Ethernet frame, an MPLS packet, or an Operations, Administration and Maintenance packet. The BIER header contains a Next Protocol field identifying the type of the payload. If the payload is an MPLS packet, an MPLS label stack immediately follows the BIER header, and the top label of that stack may be either a downstream-assigned or an upstream-assigned label.

Each router in a BIER network gets a number that is unique inside its sub-domain. A packet's header carries a row of bits, one per router, and the routers whose bits are set are the ones that must receive a copy.

§2The BIER Header

The BIER header begins with the BIFT-id, which represents a particular Bit Index Forwarding Table. Each such table corresponds to a particular combination of sub-domain, BitStringLength and Set Identifier. The default BitStringLength value for the encapsulations defined in the document is 256.

In an MPLS network the first four octets of the header carry the Traffic Class field, the S bit and the TTL field, as in any other MPLS label stack entry. The remaining fields follow. The Nibble field is set to the binary value 0101 in an MPLS network, which ensures that the MPLS equal-cost multipath logic will not confuse the remainder of the BIER header with an IP header or with the header of a pseudowire packet; a BFR that receives a BIER packet with any other value in the first nibble after the label stack SHOULD discard the packet and log an error. In a non-MPLS network the Nibble field SHOULD be set to 0000 upon transmission but MUST be ignored upon reception.

The Ver field is a 4-bit field identifying the version of the BIER header, and the document specifies version 0. If a packet is received by a BFR that does not support the specified version, the BFR MUST discard the packet and log an error. The value 0xF is reserved for experimental use and MUST NOT be assigned by any future IETF document or by IANA.

The BSL field is a 4-bit field encoding the length in bits of the BitString. When parsing the BIER header, a BFR MUST infer the length of the BitString from the BIFT-id and MUST NOT infer it from this field, which is present only to enable offline tools such as LAN analysers to parse the header. Only certain values are supported: 1 for 64 bits, 2 for 128 bits, 3 for 256 bits, 4 for 512 bits, 5 for 1024 bits, 6 for 2048 bits and 7 for 4096 bits. A received packet with any other value SHOULD be discarded and an error logged.

The Entropy field is a 20-bit field specifying a value that can be used for load-balancing purposes. The BIER forwarding process may do equal-cost load balancing, in which case the procedure MUST choose the same path for any two packets that have the same entropy value and the same BitString.

The OAM bits are set to 0 by the ingress router by default and are not modified by other routers; they have no effect on the path taken or on the quality of service applied. The Rsv bits are currently unused, SHOULD be set to 0 upon transmission and MUST be ignored upon reception. The DSCP field is not used by default in MPLS networks, and all six bits SHOULD be set to 0 upon transmission and ignored upon reception; in non-MPLS networks it MAY hold a Differentiated Services Codepoint.

The Proto field is a 6-bit Next Protocol field identifying the type of the payload, populated from the 'BIER Next Protocol Identifiers' registry. If an egress router receives a BIER packet but does not recognise or support the value, it SHOULD discard the packet and log an error. The BFIR-id field is, by default, the BFR-id of the ingress router in the sub-domain to which the packet has been assigned, encoded as an unsigned integer in the range [1,65535].

The BitString holds the bits that, together with the packet's Set Identifier and sub-domain, identify the destination egress routers. The Set Identifier and sub-domain are not carried explicitly in the BIER header, because a particular BIFT-id always corresponds to a particular Set Identifier and sub-domain.

The header has a fixed layout: which forwarding table to use, how long the packet may live, what the payload is, and the row of bits naming the destination routers.

THE BIER HEADER (RFC 8296, FIGURE 1)
BIER header
FieldNameSizeValueDescription
BIFT-IDBIFT-id20 bitsIdentifies the forwarding table to use. In an MPLS network this field carries the BIER-MPLS label.
TCTC3 bitsThe usual MPLS Traffic Class field.
S-BITS1 bitsThe MPLS bottom-of-stack bit. It must be 1 for the BIER label and 0 for any label before it.
TTLTTL1 byteThe MPLS Time to Live field, used to stop BIER packets looping.
NIBBLENibble4 bitsSet to 0101 in an MPLS network so that equal-cost multipath logic does not mistake the header for an IP or pseudowire header.
VERVer4 bitsThe version of the BIER header. This document specifies version 0.
BSLBSL4 bitsThe length in bits of the BitString. Parsers must infer the length from the BIFT-id instead.
ENTROPYEntropy20 bitsA value for load-balancing. Packets with the same entropy and the same BitString take the same path.
OAMOAM2 bitsSet to 0 by default and not modified by other routers.
RSVRsv2 bitsCurrently unused. Should be set to 0 on transmission and ignored on reception.
DSCPDSCP6 bitsUnused by default in MPLS networks; may hold a Differentiated Services Codepoint in non-MPLS networks.
PROTOProto6 bitsIdentifies the type of the payload that follows the header.
BFIR-IDBFIR-id2 bytesThe BFR-id of the ingress router, in the range [1,65535].
BITSTRINGBitString32 bytesOne bit per egress router. A set bit means that router must receive the packet.

§2.1Encapsulation in MPLS Networks

When a BIER domain is both an IGP domain and an MPLS network, each router is assumed to use IGP extensions to advertise a set of BIER-MPLS labels. When the domain contains a single sub-domain, a router needs to advertise one label for each combination of Set Identifier and BitStringLength; if the domain contains multiple sub-domains, it needs one label per Set Identifier per BitStringLength for each sub-domain. In some environments the only routing protocol in a BIER domain might be BGP, in which case BGP extensions can be used to advertise the necessary labels.

The BIER-MPLS labels are locally significant, that is, unique only to the router that advertises them, downstream-assigned MPLS labels. Penultimate hop popping MUST NOT be applied to a BIER-MPLS label. The label corresponding to a particular combination of sub-domain, Set Identifier and BitStringLength is interpreted as representing the forwarding table that corresponds to that same combination, so the label performs the function of a BIFT-id, and this label value is carried in the BIFT-id field of the encapsulation.

It is crucial that in an MPLS network the first four octets of the BIER encapsulation header are also the last four octets of the MPLS header, so any prior MPLS label stack entries MUST have the S bit clear. The S bit of the BIER-MPLS label itself MUST be set to 1. The TTL field is the usual MPLS Time to Live field, and when a BIER packet is forwarded to one or more router adjacencies, the BIER-MPLS label carried by the forwarded packet MUST have a TTL field whose value is one less than that of the packet's incoming TTL.

If a BIER packet's incoming TTL is 1 or greater and one of the bits in its BitString identifies the current router, then that router is an egress router for the packet and MUST process it as one, removing the BIER encapsulation and processing the payload based on the contents of the Proto field. If the incoming TTL is 0 the packet is considered expired, and if the incoming TTL is 1 and the BitString has a bit set that does not identify the current router the packet is also considered expired. Expired packets SHOULD be passed to an error-handling procedure, and optional implementation-specific rate limiting may be applied.

Note that if a received BIER packet has an incoming TTL of 1 and its BitString has a bit set identifying the current router, the payload MUST be processed by that router, but the packet MUST NOT be forwarded further, and it SHOULD also be passed to the error-handling procedures for expired packets.

In an MPLS network the first four bytes of the BIER header double as an ordinary MPLS label, so the packet travels like any other labelled packet until a router recognises its own label.

§2.2Encapsulation in Non-MPLS Networks

In non-MPLS networks, a BIFT-id MUST be assigned for every combination of sub-domain, Set Identifier and BitStringLength that is to be used in that network. The correspondence between a BIFT-id and a particular triple is unique throughout the BIER domain and is known to all the routers in it. The means by which the BIFT-ids are assigned and made known are outside the scope of the document. In an MPLS network the BIFT-id is an MPLS label, so its value may change as a BIER packet goes from router to router; in a non-MPLS network, since the BIFT-id is domain-wide unique, it is not expected to change as a BIER packet travels.

By default the Traffic Class field has no significance in a non-MPLS network, and should be set to the binary value 000 upon transmission and ignored upon reception. The S bit has no significance either; it should be set to 1 upon transmission and ignored upon reception. The TTL field is the BIER Time to Live field, whose purpose is to prevent BIER packets from looping indefinitely in the event of improper operation of the control plane.

A BIER packet may itself need to be carried inside another encapsulation, for example an Ethernet frame or a UDP packet, in which case the BIER packet is the payload of the outer encapsulation. The document assumes that the frame or packet carrying a BIER packet as its payload is a unicast frame or packet: although a BIER packet is a multicast packet, the frame carrying it is unicast from one router to the next.

In an MPLS network the outer encapsulation's next protocol codepoint for MPLS MUST be used, and if the outer encapsulation has a codepoint for MPLS with a downstream-assigned label and a different one for MPLS with an upstream-assigned label, the codepoint for MPLS with a downstream-assigned label MUST be used. For example, if a BIER packet is encapsulated in an Ethernet frame, the Ethertype MUST be 0x8847.

In a non-MPLS network the outer encapsulation's next protocol codepoint MUST be set to a value that means 'non-MPLS BIER', and a codepoint that means MPLS, with either upstream-assigned or downstream-assigned labels, MUST NOT be used. By requiring a distinct codepoint, non-MPLS BIER can coexist with non-BIER MPLS, and the BIFT-id values used by the former will not conflict with MPLS label values used by the latter. Therefore, if a non-MPLS BIER packet is encapsulated in an Ethernet header, the Ethertype MUST NOT be 0x8847 or 0x8848. IEEE has assigned Ethertype 0xAB37 for non-MPLS BIER packets.

Outside MPLS there is no label to carry the table number, so a number that means the same thing everywhere in the network is used instead, and a separate Ethernet type marks the packet as BIER.

THE SAME HEADER IN TWO KINDS OF NETWORK (RFC 8296, SECTION 2)
FieldIn an MPLS networkIn a non-MPLS network
BIFT-id THIS RFCThe BIER-MPLS label, which may change from router to router.A domain-wide unique number, which is not expected to change.
S bitMust be 1 for the BIER label and 0 for any label stack entry before it.No significance. Should be set to 1 and ignored on reception.
NibbleSet to 0101 so the equal-cost multipath logic does not misread the header.Should be set to 0000 and ignored on reception.
DSCPNot used by default. All six bits set to 0 on transmission.May hold a Differentiated Services Codepoint.
Ethernet type of an outer frame0x8847 for MPLS with a downstream-assigned label.0xAB37, and never 0x8847 or 0x8848.

§3Imposing and Processing the Encapsulation

Each ingress router is expected to know the Maximum Transmission Unit of the BIER domain and also the size of the BIER encapsulation, so it can deduce the maximum size of the payload that can be encapsulated in a BIER packet, a size the document calls the BIER-MTU. If an ingress router receives a multicast packet from outside the BIER domain and the packet size exceeds the BIER-MTU, it takes whatever action is appropriate for a multicast packet that is too large to be forwarded to all its next hops: if that is to drop the packet and advertise an MTU to the source, it drops the packet and advertises the BIER-MTU; if the appropriate action is to fragment the packet, the procedures below are applied to each fragment in sequence.

The ingress router then consults the multicast flow overlay to determine the value of the Proto field and the set of egress routers that must receive the packet. If more than one sub-domain is supported, it assigns the packet to a particular sub-domain. It looks up the BFR-id, in the given sub-domain, of each of the egress routers, and converts each such BFR-id into Set Identifier and BitString format.

All the BFR-ids that have the same Set Identifier can be encoded into the same BitString. For each distinct Set Identifier that occurs in the list of the packet's destination egress routers, the ingress router makes a copy of the multicast data packet and encapsulates the copy in a BIER header. The header contains the BitString representing all the destination egress routers whose BFR-ids, in the given sub-domain, correspond to the given Set Identifier, and it also contains the ingress router's own BFR-id in the sub-domain to which the packet has been assigned. The ingress router then applies the BIER forwarding procedure to that copy, which may result in one or more copies of the packet, possibly with a modified BitString, being transmitted to a neighbouring router.

If the non-MPLS BIER encapsulation is being used, the BIFT-id field is set to the BIFT-id that corresponds to the packet's sub-domain, Set Identifier and BitStringLength, and the TTL is set according to policy. If the MPLS BIER encapsulation is being used, the ingress router finds the BIER-MPLS label that the neighbour advertised as corresponding to the given combination, and prepends a label stack containing that one label, with the S bit set to indicate the end of the stack and the TTL set according to policy.

When an intermediate router processes a received MPLS packet and one of its own BIER-MPLS labels rises to the top of the label stack, the router infers the BitStringLength from the label, and the Set Identifier and sub-domain are implicitly identified by it as well. If it forwards a copy of the packet to a neighbouring router, it first swaps the label at the top of the stack with the BIER-MPLS label that the neighbour advertised for the same combination, and the TTL field of the outgoing BIER-MPLS label MUST be one less than the incoming TTL. When an intermediate router processes a received non-MPLS BIER packet, it infers the BitStringLength from the BIFT-id and the Set Identifier and sub-domain are likewise implicit, and it then follows the BIER forwarding procedures.

The ingress router works out which routers must receive the packet, turns their numbers into bits, and stamps the header. Each router along the way reads its own bit, forwards the packet on with the right label, and copies it where the bits say more copies are needed.

WHAT THE INGRESS ROUTER DOES (RFC 8296, SECTION 3)
  • Work out the destinationsConsult the multicast flow overlay for the payload type and for the set of egress routers that must receive the packet.
  • Choose a sub-domainIf more than one sub-domain is supported, the packet is assigned to one of them.
  • Turn numbers into bitsLook up the BFR-id of each destination router and convert each into Set Identifier and BitString form.
  • One copy per set identifierFor each distinct Set Identifier, a copy of the packet is made and wrapped in a BIER header carrying the matching BitString.
  • Label it for the next hopIn an MPLS network the BIER-MPLS label for that combination is pushed on the stack, with the S bit set and the TTL set by policy.

§4IANA Considerations

IANA has set up a registry called 'BIER Next Protocol Identifiers', with a registration policy of IETF Review. The initial values are 0 for Reserved, 1 for an MPLS packet with a downstream-assigned label at the top of the stack, 2 for an MPLS packet with an upstream-assigned label at the top of the stack, 3 for an Ethernet frame, 4 for an IPv4 packet, 5 for an OAM packet, 6 for an IPv6 packet, and 63 for Reserved. IEEE has assigned Ethertype 0xAB37 for non-MPLS BIER packets.

IANA keeps the list of payload types that can follow the header, and IEEE has set aside an Ethernet type for the non-MPLS form.

THE BIER NEXT PROTOCOL IDENTIFIERS (RFC 8296, SECTION 4)
  • 0Reserved — Reserved.
  • 1MPLS, downstream-assigned label — MPLS packet with a downstream-assigned label at the top of the stack.
  • 2MPLS, upstream-assigned label — MPLS packet with an upstream-assigned label at the top of the stack.
  • 3Ethernet frame — The payload is an Ethernet frame.
  • 4IPv4 packet — The payload is an IPv4 packet.
  • 5OAM packet — The payload is an Operations, Administration and Maintenance packet.
  • 6IPv6 packet — The payload is an IPv6 packet.
  • 63Reserved — Reserved.

§6Security Considerations

Insofar as the document makes use of MPLS, it inherits any security considerations that apply to the use of the MPLS data plane. If a BIER encapsulation header is modified in ways other than those specified, packets may be lost, stolen or otherwise misdelivered. Such modifications are likely to go undetected, because the BIER encapsulation does not provide cryptographic integrity protection.

Layer 2 encryption can be used to ensure that a BIER-encapsulated packet is not altered while in transit between adjacent routers. If a router itself is compromised, there is no way to prevent the compromised router from sending incorrect packets. BIER-encapsulated packets should generally not be accepted from untrusted interfaces or tunnels: an operator may wish to have a policy of accepting them only from interfaces to trusted routers and not from customer-facing interfaces.

There may be applications that require a router to accept a BIER-encapsulated packet from an interface to a system that is not controlled by the network operator, for instance a virtual machine in a data centre that submits BIER-encapsulated packets to a router. In such a case it is desirable to verify that the packet is from a legitimate source and that its BitString denotes only systems to which that source is allowed to send. The BIER encapsulation itself provides no way to verify that the source is legitimate, that it really is the system denoted by the BFIR-id, or that it is allowed to set any particular set of bits in the BitString. Insofar as the document relies on IGP extensions, it inherits any security considerations that apply to the IGP, and the security considerations of the BIER architecture apply as well.

The header carries no protection of its own, so a change to it would go unnoticed. Encryption between neighbouring routers is the way to guard against that.

§8References

[RFC2119]RFC 2119Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.March 1997[RFC2474]RFC 2474Nichols, K., Blake, S., Baker, F., and D. Black, "Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers", RFC 2474, December 1998.December 1998[RFC3031]RFC 3031Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol Label Switching Architecture", RFC 3031, January 2001.January 2001[RFC3032]RFC 3032Rosen, E., Tappan, D., Fedorkow, G., Rekhter, Y., Farinacci, D., Li, T., and A. Conta, "MPLS Label Stack Encoding", RFC 3032, January 2001.January 2001[RFC5331]RFC 5331Aggarwal, R., Rekhter, Y., and E. Rosen, "MPLS Upstream Label Assignment and Context-Specific Label Space", RFC 5331, August 2008.August 2008[RFC5332]RFC 5332Eckert, T., Rosen, E., Ed., Aggarwal, R., and Y. Rekhter, "MPLS Multicast Encapsulations", RFC 5332, August 2008.August 2008[RFC5462]RFC 5462Andersson, L. and R. Asati, "Multiprotocol Label Switching (MPLS) Label Stack Entry: EXP Field Renamed to Traffic Class Field", RFC 5462, February 2009.February 2009[RFC6790]RFC 6790Kompella, K., Drake, J., Amante, S., Henderickx, W., and L. Yong, "The Use of Entropy Labels in MPLS Forwarding", RFC 6790, November 2012.November 2012[RFC8126]RFC 8126Cotton, M., Leiba, B., and T. Narten, "Guidelines for Writing an IANA Considerations Section in RFCs", BCP 26, RFC 8126, June 2017.June 2017[RFC8279]RFC 8279Wijnands, IJ., Ed., Rosen, E., Ed., Dolganow, A., Przygienda, T., and S. Aldrin, "Multicast Using Bit Index Explicit Replication (BIER)", RFC 8279, November 2017.November 2017