Network Working Group                                      Yakov Rekhter
Internet Draft                                             Eric C. Rosen                                          Juniper Networks
Expiration Date: July 2000 2001
                                                           Eric C. Rosen
                                                     Cisco Systems, Inc.

                                                            January 2000 2001

                  Carrying Label Information in BGP-4



Status of this Memo

   This document is an Internet-Draft and is in full conformance with
   all provisions of Section 10 of RFC2026.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF), its areas, and its working groups.  Note that
   other groups may also distribute working documents as Internet-

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   The list of current Internet-Drafts can be accessed at

   The list of Internet-Draft Shadow Directories can be accessed at


   When BGP is used to distribute a particular route, it can be also be
   used to distribute an MPLS label which is mapped to that route
   [MPLS-ARCH].  This document specifies the way in which this is done.
   The label mapping information for a particular route is piggybacked
   in the same BGP Update message that is used to distribute the route

Table of Contents

    1      Specification of Requirements  ..........................   2
    2      Overview  ...............................................   2
    3      Carrying Label Mapping Information  .....................   3
    4      Advertising Multiple Routes to a Destination  ...........   4
    5      Capability Negotiation  .................................   5
    6      When the BGP Peers are not Directly Adjacent  ...........   5
    7      Security Considerations  ................................   6
    8      Acknowledgments  ........................................   7
    9      References  .............................................   7
   10      Author Information  .....................................   7

1. Specification of Requirements

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

2. Overview

   When BGP is used to distribute a particular route, it can also be
   used to distribute an MPLS label that is mapped to that route [MPLS-
   ARCH].  This document specifies the way in which this is done.  The
   label mapping information for a particular route is piggybacked in
   the same BGP Update message that is used to distribute the route

   This can be useful in the following situations:

     - If two immediately adjacent Label Switched Routers (LSRs) are
       also BGP peers, then label distribution can be done without the
       need for any other label distribution protocol.

     - Suppose one's network consists of two "classes" of LSR: exterior
       LSRs, which interface to other networks, and interior LSRs, which
       serve only to carry traffic between exterior LSRs.  Suppose that
       the exterior LSRs are BGP speakers. If the BGP speakers
       distribute MPLS labels to each other along with each route they
       distribute, then as long as the interior routers support MPLS,
       they need not receive any of the BGP routes from the BGP

       If exterior router A needs to send a packet to destination D, and
       A's BGP next hop for D is exterior router B, and B has mapped
       label L to D, then A first pushes L onto the packet's label
       stack.  A then consults its IGP to find the next hop to B, call
       it C.  If C has distributed to A an MPLS label for the route to
       B, A can push this label on the packet's label stack, and then
       send the packet to C.

   If a set of BGP speakers are exchanging routes via a Route Reflector
   [BGP-RR], then by piggybacking the label distribution on the route
   distribution, one is able to use the Route Reflector to distribute
   the labels as well.  This improves scalability quite significantly.
   Note that if the Route Reflector is not in the forwarding path, it
   need not even be capable of forwarding MPLS packets.

   Label distribution can be piggybacked in the BGP Update message by
   using the BGP-4 Multiprotocol Extensions attribute [RFC 2283].  The
   label is encoded into the NLRI field of the attribute, and the SAFI
   ("Subsequent Address Family Identifier") field is used to indicate
   that the NLRI contains a label.  A BGP speaker may not use BGP to
   send labels to a particular BGP peer unless that peer indicates,
   through BGP Capability Negotiation, that it can process Update
   messages with the specified SAFI field.

3. Carrying Label Mapping Information

   Label mapping information is carried as part of the Network Layer
   Reachability Information (NLRI) in the Multiprotocol Extensions
   attributes.  The AFI indicates, as usual, the address family of the
   associated route.  The fact that the NLRI contains a label is
   indicated by using SAFI value 4 [assignment to be confirmed by IANA].

   The Network Layer Reachability information is encoded as one or more
   triples of the form <label, length, <length, label, prefix>, whose fields are
   described below:

      |   Length (1 octet)        |
      |   Label (3 octets)        |
      |   Prefix (variable)       |
   The use and the meaning of these fields are as follows:

      a) Length:

         The Length field indicates the length in bits of the address
         prefix plus the label(s).

      b) Label:

         The Label field carries one or more labels (that corresponds to
         the stack of labels [MPLS-ENCAPS]). Each label is encoded as 3
         octets, where the high-order 20 bits contain the label value,
         and the low order bit contains "Bottom of Stack" (as defined in
         [MPLS-ENCAPS]). The following high-order three bits
         must be zero.  The remaining 20 bits contain the label value.

      c) Prefix:

         The Prefix field contains address prefixes followed by enough
         trailing bits to make the end of the field fall on an octet
         boundary.  Note that the value of trailing bits is irrelevant.

   The label(s) specified for a particular route (and associated with it
   its address prefix) must be assigned by the LSR which is identified
   by the value of the Next Hop attribute of the route.

   When a BGP speaker redistributes a route, the label(s) assigned to
   that route must not be changed (except by omission), unless the
   speaker changes the value of the Next Hop attribute of the route.

   A BGP speaker can withdraw a previously advertised route (as well as
   the binding between this route and a label) by either (a) advertising
   a new route (and a label) with the same NLRI as the previously
   advertised route, or (b) listing the NLRI of the previously
   advertised route in the Withdrawn Routes field of an Update message.
   The label information carried (as part of NLRI) in the Withdrawn
   Routes field should be set to 0x800000.  (Of course, terminating the
   BGP session also withdraws all the previously advertised routes.)

4. Advertising Multiple Routes to a Destination

   A BGP speaker may maintain (and advertise to its peers) more than one
   route to a given destination, as long as each such route has its own

   The encoding described above allows a single BGP Update message to
   carry multiple routes, each with its own label(s).

   In the case where a BGP speaker advertises multiple routes to a
   destination, if a route is withdrawn, and a label(s) is specified at
   the time of withdrawal, only the corresponding route with the
   corresponding label is withdrawn. If a route is withdrawn, and no
   label is specified at the time of withdrawal, then only the
   corresponding unlabeled route is withdrawn; the labeled routes are
   left in place.

5. Capability Negotiation

   A BGP speaker that uses Multiprotocol Extensions to carry label
   mapping information should use the Capabilities Optional Parameter,
   as defined in [BGP-CAP], to inform its peers about this capability.
   The MP_EXT Capability Code, as defined in [BGP-MP], is used to
   negotiate the (AFI, SAFI) pairs available on a particular connection.

   A BGP speaker should not advertise this capability to another BGP
   speaker unless there is a Label Switched Path (LSP) between the two

   A BGP speaker that is capable of handling multiple routes to a
   destination (as described above) should use the Capabilities Optional
   Parameter, as defined in [BGP-CAP], to inform its peers about this
   capability. The value of this capability is TBD.

6. When the BGP Peers are not Directly Adjacent

   Consider the following LSR topology: A--B--C--D.  Suppose that D
   distributes a label L to A.  In this topology, A cannot simply push L
   onto a packet's label stack, and then send the resulting packet to B.
   D must be the only LSR that sees L at the top of the stack.  Before A
   sends the packet to B, it must push on another label, which was
   distributed by B.  B must replace this label with yet another label,
   which was distributed by C.  In other words, there must be an LSP
   between A and D.  If there is no such LSP, A cannot make use of label
   L.  This is true any time labels are distributed between non-adjacent
   LSRs, whether that distribution is done by BGP or by some other

   This document does NOT specify any procedure for ensuring in real
   time that label distribution between non-adjacent LSRs is done only
   when the appropriate MPLS infrastructure exists in the network or
   networks connecting the two LSRs.  Ensuring that the proper
   infrastructure exists is an issue for network management and

7. Security Considerations

   When an LSR A is directly connected to an LSR B via a point-to-point
   interface, then when A receives packets over that interface, it knows
   that they come from B.  This makes it easy for A to discard any
   packets from B whose top labels are not among the labels that A
   distributed to B.  That is, A can easily ensure that B only uses
   those labels which it is entitled to use.  This technique can be used
   to prevent "label spoofing", i.e., the situation in which an LSR
   imposes a label which has not been properly distributed to it.

   The procedures discussed in this document would commonly be used when
   the label distribution peers are separated not merely by a point-to-
   point link, but by an MPLS network.  This means that when an LSR A
   processes a labelled packet, it really has no way to determine which
   other LSR B pushed on the top label.  Hence it cannot tell whether
   the label is one which B is entitled to use.  In fact, when Route
   Reflectors are in use, A may not even know the set of LSRs which
   receive its label mappings.  So the previous paragraph's technique
   for preventing label spoofing does not apply.

   It is possible though to use other techniques to avoid label spoofing
   problems.  If, for example, one never accepts labeled packets from
   the network's "external" interfaces, and all the BGP-distributed
   labels are advertised via IBGP, then there is no way for an untrusted
   router to put a labeled packet into the network.  One can generally
   assume that one's IBGP peers (or the IBGP peers of one's Route
   Reflector) will not attempt label spoofing, since they are all under
   the control of a single administration.

   This condition can actually be weakened significantly.  One doesn't
   need to refuse to accept all labeled packets from external
   interfaces.  One just needs to make sure that any labeled packet
   received on an external interface has a top label which was actually
   distributed out that interface.

   Then a label spoofing problem would only exist if there are both
   trusted and untrusted systems out the same interface.  One way to
   avoid this problem is simply to avoid this situation.

8. Acknowledgments

   Thanks to Ravi Chandra, Enke Chen, Srihari Ramachandra, Eric Gray and
   Liam Casey for their comments.

9. References

   [BGP-4] RFC 1771, "A Border Gateway Protocol 4 (BGP-4)", Y. Rekhter,
   T. Li, 3/95

   [BGP-CAP] "Capabilities Negotiation with BGP-4", R. Chandra, J.
   Scudder, draft-ietf-idr-bgp4-cap-neg-04.txt, 9/99

   [BGP-MP] RFC 2283, "Multiprotocol Extensions for BGP-4", T. Bates, R.
   Chandra, D.Katz, Y. Rekhter, 2/98

   [BGP-RR] RFC 1966, "BGP Route Reflection: An alternative to full mesh
   IBGP", T. Bates, R. Chandra, 6/96.

   [MPLS-ARCH] "Multiprotocol Label Switching Architecture"A Proposed
   Architecture for MPLS", E. Rosen, A. Vishwanathan, R. Callon, draft-
   ietf-mpls-arch-06.txt, 8/99.

   [MPLS-ENCAPS] "MPLS Label Stack Encoding", E. Rosen, et al, draft-
   ietf-mpls-label-encaps-07.txt, 9/99

10. Author Information

   Yakov Rekhter
   Cisco Systems, Inc.
   170 West Tasman Drive
   San Jose,
   Juniper Networks
   1194 N. Mathilda Avenue
   Sunnyvale, CA 95134 94089

   Eric Rosen
   Cisco Systems, Inc.
   250 Apollo Drive
   Chelmsford, MA 01824