Network Working Group                                           O. Troan
Internet-Draft                                             Cisco Systems
Intended status: Standards Track                      September 28, 2018
Expires: April 1, 2019

                   Path MTU discovery solution space


   Path MTU discovery has turned out to be a thorny problem that has
   haunted the Internet community for decades.  Lately there has been
   some work both at the transport layer and at the network layer.  This
   memo lists the solutions the author is aware of from the perspective
   of the network layer.

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at

   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."

   This Internet-Draft will expire on April 1, 2019.

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
   ( in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Troan                     Expires April 1, 2019                 [Page 1]

Internet-Draft                                            September 2018

1.  Introduction

   Path MTU discovery has turned out to be harder than expeced.  In IPv6
   we set out following the same model as for IPv4.  The sending host
   maintains a MTU cache, that is updated based on received ICMP PMTUD
   messages.  That solution has a few short-comings:

   o  Sending of ICMP PMTUD messages is throttled in routers [RFC4443]

   o  It's not efficient if links along the path have decreasingly
      smaller MTU, then multiple rounds of large packet, resulting ICMP
      PMTUD happens.

   o  ICMP might be ignored by host stacks / applications

   o  As ICMP looks different than application traffic, it might be
      blocked by routers.

   o  Doesn't work well in an anycast scenario (but what does).

2.  Requirements / Goals

   1.  Avoid MTU black-holes [RFC2923].

   2.  Detect the Path MTU in single round trip.

   3.  Adapt to varying MTU over the connection life time.

   4.  The signalling of the MTU back to the sender must be
       indistinguishable from application traffic to lessen risk of

   5.  Design a mechanism that ensures that neither MTU probes nor MTU
       signalling back to sender are more likely to be dropped than
       other application traffic.

   6.  Must be deployable and anchored in transport / application areas.

   7.  [Optional?]  Support neighbors on the same link which support
       higher MTU than link MTU see [I-D.van-beijnum-multi-mtu]

3.  Network layer solutions for Path MTU discovery

   o  PMTUD [RFC8201]

   o  On-path fragmentation, IPv4 style.  We know this one.

Troan                     Expires April 1, 2019                 [Page 2]

Internet-Draft                                            September 2018

   o  Packet truncation.  [I-D.leddy-6man-truncate].  The source sets a
      truncation elligble flag in the packet, routers on the path may
      truncate f the packet is too big, and sets a truncated done flag.
      Then the receiver signals the learnt forward MTU back to the
      sender.  Either via existing ICMP PMTUD or a transport layer
      option.  This is an example of a solution which does not require
      the sender having to accept packets from intermediate nodes.

   o  MTU recording.  Probe packets are sent, either as part of data
      packets, if those are guaranteed not to exceed MTU.  Some trigger
      in the header (ECN like flags) or a HBH option is required for the
      router to record the smallest MTU along the path.  Application /
      Transport would have to periodically include the probe trigger in
      data packets to detect changes in path MTU.

3.1.  Common problems

   How is the router along the path "triggered" to put this packet on
   the exception path?  For current and the truncation scheme it's a
   simple check in the forwarding path for the size of packet versus
   outgoing interface MTU.  For e.g. a recording MTU mechanism it would
   have to be flags in the IPv6 header or an HBH option.

   How should the forward path MTU be signalled back to the sender?  The
   signal should look like any other application traffic to avoid
   filtering or is it sufficient to avoid sending from intermitent

4.  Solutions at other layers

   In addition there are solutions at the transport layer, that work in
   co-hort or independently of the network layer soltusions.  [RFC4821]
   and [I-D.ietf-tsvwg-datagram-plpmtud].

   One could also imagine other solutions, e.g. to include MTU in router
   advertisements in BGP, so that a BGP speaker could calculate the end
   to end MTU across the set of administrative domains.

5.  Conclusion

   What are our options?  Even if we developed a new PMTU mechanism, IP
   stacks must deal with networks where the new mechanism isn't yet
   deployed.  Will a new mechanism be so much better that it provides
   enough value for it to be deployed?  Or should we at the network
   layer just punt this to transport?

Troan                     Expires April 1, 2019                 [Page 3]

Internet-Draft                                            September 2018

6.  References

              Fairhurst, G., Jones, T., Tuexen, M., and I. Ruengeler,
              "Packetization Layer Path MTU Discovery for Datagram
              Transports", draft-ietf-tsvwg-datagram-plpmtud-04 (work in
              progress), September 2018.

              Leddy, J. and R. Bonica, "IPv6 Packet Truncation", draft-
              leddy-6man-truncate-04 (work in progress), June 2018.

              Beijnum, I., "Extensions for Multi-MTU Subnets", draft-
              van-beijnum-multi-mtu-05 (work in progress), March 2016.

   [RFC2923]  Lahey, K., "TCP Problems with Path MTU Discovery",
              RFC 2923, DOI 10.17487/RFC2923, September 2000,

   [RFC4443]  Conta, A., Deering, S., and M. Gupta, Ed., "Internet
              Control Message Protocol (ICMPv6) for the Internet
              Protocol Version 6 (IPv6) Specification", STD 89,
              RFC 4443, DOI 10.17487/RFC4443, March 2006,

   [RFC4821]  Mathis, M. and J. Heffner, "Packetization Layer Path MTU
              Discovery", RFC 4821, DOI 10.17487/RFC4821, March 2007,

   [RFC8201]  McCann, J., Deering, S., Mogul, J., and R. Hinden, Ed.,
              "Path MTU Discovery for IP version 6", STD 87, RFC 8201,
              DOI 10.17487/RFC8201, July 2017,

Author's Address

   Ole Troan
   Cisco Systems
   Philip Pedersens vei 1
   Lysaker  1366


Troan                     Expires April 1, 2019                 [Page 4]