Network Working Group                                           B. Black
Internet Draft                                           Layer8 Networks
Updates: 3036                                                K. Kompella
Category: Standards Track                               Juniper Networks
Expires: December 2003                                         June April 2004                                         October 2003

                   MTU

            Maximum Transmission Unit Signalling Extensions
                  for LDP
               draft-ietf-mpls-ldp-mtu-extensions-01.txt the Label Distribution Protocol
               draft-ietf-mpls-ldp-mtu-extensions-02.txt

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-
   Drafts.

   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
              http://www.ietf.org/ietf/1id-abstracts.txt

   The list of Internet-Draft Shadow Directories can be accessed at
              http://www.ietf.org/shadow.html.

Copyright Notice

   Copyright (C) The Internet Society (2003).  All Rights Reserved.

Abstract

   Proper functioning of RFC 1191 path MTU Maximum Transmission Unit (MTU)
   discovery requires that IP routers have knowledge of the MTU for each
   link to which they are connected.  As currently specified, the Label
   Distribution Protocol (LDP) does not have the ability to signal the
   MTU for a Label Switched Path (LSP) to the ingress Label Switching
   Router (LSR).  In the absence of this functionality, the MTU for each
   LSP must be statically configured by network operators or by
   equivalent, off-line mechanisms.

   This document specifies extensions to LDP in support of LSP MTU
   discovery.

Conventions used in this document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED",  "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC-2119 [1].

Changes from last version

   [Note to RFC Editor: please remove this section before publishing.]

   Incorporated suggestions received on the MPLS WG mailing list, mostly
   clarifications and editorial nits.

   Expanded examples.  Made slight formatting changes.

   Updated references.

1. Introduction

   As currently specified in [2], the LDP protocol for MPLS does not
   support signalling of the MTU for LSPs to ingress LSRs.  This
   functionality is essential to the proper functioning of RFC 1191 path
   MTU detection [3].  Without knowledge of the MTU for an LSP, edge
   LSRs may transmit packets along that LSP which are, according to [4],
   too big.  Such packets may be silently discarded by LSRs along the
   LSP, effectively preventing communication between certain end hosts.

   The solution proposed in this document enables automatic
   determination of the MTU for an LSP with the addition of a TLV to
   carry MTU information for a FEC between adjacent LSRs in LDP Label
   Mapping messages.  This information is sufficient for a set of LSRs
   along the path followed by an LSP to discover either the exact MTU
   for that LSP, or an approximation which is no worse than could be
   generated with local information on the ingress LSR.

1.1. Changes from last version

   The biggest change, protocol-wise is that the notion of 'egress'
   interface has been removed.  The LSP MTU at the egress is now 65535.
   This has repurcussions on the processing of the MTU TLV.  Also, the
   MTU TLV now has both the U and F bits set.

   A number of definitions have been introduced to clarify the
   exposition.  Also, the examples have been changed significantly.

2. MTU Signalling

   The signalling procedure described in this document employs the
   addition of a single TLV to LDP Label Mapping messages and a simple
   algorithm for LSP MTU calculation.

2.1. Definitions

   Link MTU: the MTU of a given link.  This size includes the IP header
   and data (or other payload) and the label stack, but does not include
   any lower-level lower-layer headers.  A link may be an interface (such as
   Ethernet or Packet-over-SONET), a tunnel (such as GRE or IPsec) or an
   LSP.

   Peer LSRs: for LSR A and FEC F, this is the set of LSRs that sent a
   Label Mapping for FEC F to A.

   Downstream LSRs: for LSR A and FEC F, this is the subset of A's peer
   LSRs for FEC F to whom A will forward packets for the FEC.
   Typically, this subset is determined via the routing table.

   Hop MTU: the MTU of an LSP hop between an upstream LSR A and a
   downstream LSR B.  This size includes the IP header and data (or
   other payload) and the part of the label stack that is considered
   payload as far as this LSP goes.  It does not include any lower-level
   headers.  (Note: if there are multiple links between A and B, the Hop
   MTU is the minimum of the Hop MTU of those links used for
   forwarding.)
   LSP MTU: the MTU of an LSP from a given LSR to the egress(es), over
   each valid (forwarding) path.  This size includes the IP header and
   data (or other payload) and any part of the label stack that was
   received by the ingress LSR before it placed the packet into the LSP
   (this part of the label stack is considered part of the payload for
   this LSP).  The size does not include any lower-level headers.

2.2. Example

   Consider LSRs A-F interconnected as follows:

                 M       P
               _____ C =====
              /      |      \
     A ~~~~~ B ===== D ----- E ----- F
         L       N       Q       R

   Say that the link MTU for link L is 9216, for links M, Q and R is
   4470, and for N and P is 1500.

   Consider a FEC X for which F is the egress, and say that all LSRs
   advertise X to their neighbors.

   Note that while LDP may be running on the C-D link, it is not used
   for forwarding (e.g., because it has a high metric).  In particular,
   D is an LDP neighbor of C, but D is not one of C's downstream LSRs
   for FEC X.

   E's peers for FEC X are C, D and F.  Say E chooses F as its
   downstream LSR for X.  E's Hop MTU for link R is 4466.  If F
   advertised an implicit null label to E, then E MAY set the Hop MTU
   for R to 4470.

   C's peers for FEC X are B, D and E.  Say C chooses E as its
   downstream LSR for X.  Similarly, A chooses B, B chooses C and D, D
   (equal cost multi-path), D chooses E and E chooses F (respectively)
   as their downstream LSRs.

   C's Hop MTU to E for FEC X is 1496.  B's Hop MTU to C is 4466, and to
   D is 1496.  A's LSP MTU for FEC X is 1496.  If A has another LSP for
   FEC Y to F (learned via targetted LDP) that rides over the LSP for
   FEC X, the MTU for that LSP would be 1492.

   If B had a targetted LDP session to E E, say over which an RSVP-TE tunnel T,
   and B received a Mapping for FEC X, X over the targetted LDP session,
   then E would also be B's peer, and E may be chosen as a downstream
   LSR for B.  In that case, B's LSP MTU for FEC X would then be the
   smaller of {(T's MTU - 4), E's LSP MTU for X}.

   This memo describes how A determines its LSP MTU for FEC FECs X and Y.

2.3. Signalling Procedure

   The procedure for signalling the MTU is performed hop-by-hop by each
   LSR L along an LSP for a given FEC F.  The steps are as follows:

   1.  First, L computes the its LSP MTU for FEC F:

       A.  If L is the egress for F, L sets the LSP MTU for F to 65535.

       B.  [OPTIONAL] If L's only downstream LSR is the egress for F
           (i.e., L is a penultimate hop for F), and L receives an
           implicit null label as its Mapping for F, then L can set the
           Hop MTU for its downstream link to the link MTU instead of
           (link MTU - 4 octets).  L's LSP MTU for F is the Hop MTU.

       C.  Otherwise (L is not the egress LSR, LSR), L computes the LSP MTU
           for F as follows:

           a)  L determines its downstream neighbors LSRs for FEC F.

           b)  For each downstream neighbor LSR Z, L computes the minimum of the
               Hop MTU to Z and the LSP MTU in the MTU TLV that Z
               advertised to L.  If Z did not include the MTU TLV in its
               Label Mapping, then Z's LSP MTU is set to 65535.

           c)  L sets its LSP MTU to the minimum of the MTUs it computed
               for its downstream neighbors. LSRs.

   2.  For each LDP neighbor (direct or targetted) of L to which L
       decides to send a Mapping for FEC F, L attaches an MTU TLV with
       the LSP MTU that it computed for this FEC.  L MAY (because of
       policy or other reasons) advertise a smaller MTU than it has
       computed, but L MUST NOT advertise a larger MTU.

   3.  When a new MTU is received for FEC F from a downstream LSR, or
       the set of downstream LSRs for F changes, L returns to Step 1.
       If the newly computed LSP MTU is unchanged, L SHOULD NOT
       advertise new information to its neighbors.  Otherwise, L
       readvertises its Mappings for F to all its peers with an updated
       MTU TLV.

       This behavior is standard for attributes such as path vector and
       hop count, and the same rules apply, as specified in [2].

       If the LSP MTU decreases, L SHOULD readvertise the new MTU
       immediately; if the LSP MTU increases, L MAY hold down the
       readvertisement.

2.4. MTU TLV

   The MTU TLV encodes information on the maximum transmission unit for
   an LSP, from the advertising LSR to the egress(es) over all valid
   paths.

   The encoding for the MTU TLV is:

       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |1|1|      MTU TLV (0x0XXX)     |            Length             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |              MTU              |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   MTU

   This is a 16-bit unsigned integer that represents the MTU in octets
   for an LSP or segment of an LSP.

   Note that the U and F bits are set.  An LSR that doesn't recognize
   the MTU TLV MUST ignore it when it processes the Label Mapping
   message, and forward the TLV to its peers.  This may result in the
   incorrect computation of the LSP MTU; however, silently forwarding
   the MTU TLV preserves maximal amount of information about the LSP
   MTU.

3. Example of Operation

   Consider the example network in section 2.2.  The following table  Table 1 describes, for
   each LSR, the links to its downstream LSRs, the Hop MTU for the peer,
   the LSP MTU received from the peer, and the LSR's computed LSP MTU.

   Now consider the same network with the following changes: there is an
   LSP T from B to E, and a targetted LDP session from B to E.  B's peer
   LSRs are A, C, D and E; B's downstream LSRs are D and E; to reach E,
   B chooses to go over T.  The LSP MTU for LSP T is 1496.  This
   information is depicted in Table 2.

         LSR  |  Link  |  Hop MTU  |  Recvd MTU  |  LSP MTU
         --------------------------------------------------
          F   |    -   |    65535  |      -      |    65535
         --------------------------------------------------
          E   |    R   |     4466  |  F:  65535  |     4466
         --------------------------------------------------
          D   |    Q   |     4466  |  E:   4466  |     4466
         --------------------------------------------------
          C   |    P   |     1496  |  E:   4466  |     1496
         --------------------------------------------------
          B   |    M   |     4466  |  C:   1496  |
              |    N   |     1496  |  D:   4466  |     1496
         --------------------------------------------------
          A   |    L   |     9212  |  B:   1496  |     1496
         --------------------------------------------------

   Now consider the same network with the following changes: there is an
   LSP X from B to E, and a targetted LDP session from B to E.  B's peer
   LSRs are A, C, D and E; B's downstream LSRs are D and E; to reach E,
   B chooses to go over X.  The LSP MTU for LSP X is 1496.
                              Table 1

         LSR  |  Link  |  Hop MTU  |  Recvd MTU  |  LSP MTU
         --------------------------------------------------
          F   |    -   |    65535  |      -      |    65535
         --------------------------------------------------
          E   |    R   |     4466  |  F:  65535  |     4466
         --------------------------------------------------
          D   |    Q   |     4466  |  E:   4466  |     4466
         --------------------------------------------------
          C   |    P   |     1496  |  E:   4466  |     1496
         --------------------------------------------------
          B   |    X    T   |     1492  |  E:   4466  |
              |    N   |     1496  |  D:   4466  |     1492
         --------------------------------------------------
          A   |    L   |     9212  |  B:   1492  |     1492
         --------------------------------------------------
                              Table 2

4. Using the LSP MTU

   An ingress LSR that forwards an IP packet into an LSP whose MTU it
   knows MUST either fragment the IP packet to the LSP's MTU (if the
   Don't Fragment bit is clear) or drop the packet and respond with an
   ICMP Destination Unreachable message to the source of the packet,
   with the Code indicating "fragmentation needed and DF set", and the
   Next-Hop MTU set to the LSP MTU.  In other words, the LSR behaves as
   RFC 1191 says, except it treats the LSP as the next hop "network".

   If the payload for the LSP is not an IP packet, the LSR MUST forward
   the packet if it fits (size <= LSP MTU), and SHOULD drop it if it
   doesn't fit.

5. Protocol Interaction

5.1. Interaction With LSRs Which Do Not Support MTU Signalling

   Changes in MTU for sections of an LSP may cause intermediate LSRs to
   generate unsolicited label Mapping messages to advertise the new MTU.
   LSRs which do not support MTU signalling MUST accept these messages,
   but MAY ignore them (see Section 2.1).

5.2. Interaction with CR-LDP and RSVP-TE

   The MTU TLV can be used to discover the Path MTU of both LDP LSPs and
   CR-LDP LSPs.  This proposal is not impacted in the presence of LSPs
   created using CR-LDP, as specified in [5].

   Note that LDP/CR-LDP LSPs may tunnel through other LSPs signalled
   using LDP, CR-LDP or RSVP-TE [6]; the mechanism suggested here
   applies in all these cases, essentially by treating the tunnel LSPs
   as links.

Normative References

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

   [2]  Andersson, L., Doolan, P., Feldman, N., Fredette, A. and B.
        Thomas, "LDP Specification", RFC 3036, January 2001. 2001

   [3]  Mogul, J. and S. Deering, "Path MTU Discovery", RFC 1191,
        November 1990. 1990

   [4]  Rosen, E., Tappan, D., Federkow, G., Rekhter, Y., Farinacci, D.,
        Li, T. and A. Conta, "MPLS Label Stack Encoding", RFC 3032,
        January 2001. 2001

   [5]  Jamoussi, J., B., Ed., "Constraint-Based LSP Setup Using LDP", July 2000. RFC
        3212, January 2002

   [6]  Awduche, D., Berger, L. and D. L., Gan, D., Li, T., Srinivasan, V. and G.
        Swallow, "RSVP-TE: Extensions to RSVP for LSP Tunnels", February 2001. RFC
        3209, December 2001

Security Considerations

   This mechanism does not introduce any new weaknesses in LDP.  It is
   possible to spoof TCP packets belonging to an LDP session to
   manipulate the LSP MTU, but this sort of attack is not new LDP has mechanisms to LDP. thwart these types
   of attacks.

IANA Considerations

   A new LDP TLV Type is defined in section 2.4.  A Type has to be
   allocated by IANA; a number from the range 0x0000 - 0x3DFF is
   requested.

Acknowledgments

   We would like to thank Andre Fredette for a number of detailed
   comments on earlier versions of the signalling mechanism.  Eric Gray
   and Gray,
   Giles Heron and Mark Duffy have contributed numerous useful
   suggestions.

Authors' Addresses

   Benjamin Black
   Layer8 Networks

   EMail: ben@layer8.net

   Kireeti Kompella
   Juniper Networks
   1194 N. Mathilda Ave
   Sunnyvale, CA  94089
   US

   EMail: kireeti@juniper.net

IPR Notice

   The IETF takes no position regarding the validity or scope of any
   intellectual property or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; neither does it represent that it
   has made any effort to identify any such rights.  Information on the
   IETF's procedures with respect to rights in standards-track and
   standards-related documentation can be found in BCP-11.  Copies of
   claims of rights made available for publication and any assurances of
   licenses to be made available, or the result of an attempt made to
   obtain a general license or permission for the use of such
   proprietary rights by implementors or users of this specification can
   be obtained from the IETF Secretariat.

   The IETF invites any interested party to bring to its attention any
   copyrights, patents or patent applications, or other proprietary
   rights which may cover technology that may be required to practice
   this standard.  Please address the information to the IETF Executive
   Director.

Full Copyright Notice Statement

   Copyright (C) The Internet Society (2003).  All Rights Reserved.

   This document and translations of it may be copied and furnished to
   others, and derivative works that comment on or otherwise explain it
   or assist in its implementation may be prepared, copied, published
   and distributed, in whole or in part, without restriction of any
   kind, provided that the above copyright notice and this paragraph are
   included on all such copies and derivative works.  However, this
   document itself may not be modified in any way, such as by removing
   the copyright notice or references to the Internet Society or other
   Internet organizations, except as needed for the purpose of
   developing Internet standards in which case the procedures for
   copyrights defined in the Internet Standards process must be
   followed, or as required to translate it into languages other than
   English.

   The limited permissions granted above are perpetual and will not be
   revoked by the Internet Society or its successors or assigns.

   This document and the information contained herein is provided on an
   "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
   TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
   BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
   HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE."

Acknowledgement PURPOSE.

Acknowledgement:

   Funding for the RFC Editor function is currently provided by the
   Internet Society.