Network Working Group                                          M. Mathis
Internet-Draft                                                J. Heffner
Expires: April 26, 2006                                              PSC
                                                        October 23, 2005

                           Path MTU Discovery

Status of this Memo

   By submitting this Internet-Draft, each author represents that any
   applicable patent or other IPR claims of which he or she is aware
   have been or will be disclosed, and any of which he or she becomes
   aware will be disclosed, in accordance with Section 6 of BCP 79.

   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

   This Internet-Draft will expire on April 26, 2006.

Copyright Notice

   Copyright (C) The Internet Society (2005).


   This document describes a robust method for Path MTU Discovery that
   relies on TCP or some other Packetization Layer to probe an Internet
   path with progressively larger packets.  This method is described as
   an extension to RFC 1191 and RFC 1981, which specify ICMP based Path
   MTU Discovery for IP versions 4 and 6, respectively.

   The general strategy of the new algorithm is to start with a small
   MTU and search upward, testing successively larger MTUs by probing

Mathis & Heffner         Expires April 26, 2006                 [Page 1]

Internet-Draft             Path MTU Discovery               October 2005

   with single packets.  If the probe is successfully delivered and
   satisfies a subsequent verification phase then the MTU is raised.  If
   the probe is lost, it is treated as an MTU limitation and not as a
   congestion signal.

   There are several options for integrating PLPMTUD with classical path
   MTU discovery.  PLPMTUD can be minimally configured to perform ICMP
   black hole recovery to increase the robustness of classical path MTU
   discovery, or ICMP processing can be completely disabled, and PLPMTUD
   can completely replace classical path MTU discovery.

   In the latter configuration, PLPMTUD exactly parallels congestion
   control.  An end-to-end transport protocol adjusts non-protocol
   properties of the data stream (window size or packet size) while
   using packet losses to deduce the appropriateness of the adjustments.
   This technique seems to be more philosophically consistent with the
   end-to-end principle than relying on ICMP messages containing
   transcribed headers of multiple protocol layers.

Mathis & Heffner         Expires April 26, 2006                 [Page 2]

Internet-Draft             Path MTU Discovery               October 2005

Table of Contents

   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  4
     1.1.  Revision History . . . . . . . . . . . . . . . . . . . . .  4
       1.1.1.  Changes since version -04, February 2005 (IETF 62) . .  5
       1.1.2.  Changes since version -03, October 2004 (IETF 61)  . .  5
       1.1.3.  Changes since version -02, July 19th 2004 (IETF 60)  .  5
   2.  Overview . . . . . . . . . . . . . . . . . . . . . . . . . . .  6
   3.  Terminology  . . . . . . . . . . . . . . . . . . . . . . . . .  9
   4.  Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 11
   5.  Layering . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
     5.1.  Accounting for header sizes  . . . . . . . . . . . . . . . 13
     5.2.  Storing PMTU information . . . . . . . . . . . . . . . . . 13
     5.3.  Accounting for IPsec . . . . . . . . . . . . . . . . . . . 14
     5.4.  Multicast  . . . . . . . . . . . . . . . . . . . . . . . . 15
   6.  Common Packetization Properties  . . . . . . . . . . . . . . . 15
     6.1.  Mechanism to detect loss . . . . . . . . . . . . . . . . . 15
     6.2.  Generating probes  . . . . . . . . . . . . . . . . . . . . 16
   7.  Host Fragmentation . . . . . . . . . . . . . . . . . . . . . . 16
   8.  The Probing Method . . . . . . . . . . . . . . . . . . . . . . 17
     8.1.  Packet size ranges . . . . . . . . . . . . . . . . . . . . 17
     8.2.  Selecting initial values . . . . . . . . . . . . . . . . . 18
     8.3.  Selecting probe size . . . . . . . . . . . . . . . . . . . 19
     8.4.  Probing preconditions  . . . . . . . . . . . . . . . . . . 20
     8.5.  Conducting a probe . . . . . . . . . . . . . . . . . . . . 20
     8.6.  Response to probe results  . . . . . . . . . . . . . . . . 20
       8.6.1.  Probe success  . . . . . . . . . . . . . . . . . . . . 21
       8.6.2.  Probe failure  . . . . . . . . . . . . . . . . . . . . 21
       8.6.3.  Probe timeout failure  . . . . . . . . . . . . . . . . 21
       8.6.4.  Probe inconclusive . . . . . . . . . . . . . . . . . . 22
     8.7.  Full stop timeout  . . . . . . . . . . . . . . . . . . . . 22
     8.8.  MTU verification . . . . . . . . . . . . . . . . . . . . . 22
   9.  Diagnostic Interface . . . . . . . . . . . . . . . . . . . . . 23
   10. Specific Packetization Layers  . . . . . . . . . . . . . . . . 23
     10.1. Probing method using TCP . . . . . . . . . . . . . . . . . 24
     10.2. Probing method using SCTP  . . . . . . . . . . . . . . . . 24
     10.3. Probing method using IP fragmentation  . . . . . . . . . . 25
     10.4. Probing method using applications  . . . . . . . . . . . . 26
   11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 27
     11.1. Normative references . . . . . . . . . . . . . . . . . . . 27
     11.2. Informative references . . . . . . . . . . . . . . . . . . 28
   Appendix A.  Security Considerations . . . . . . . . . . . . . . . 29
   Appendix B.  IANA Considerations . . . . . . . . . . . . . . . . . 29
   Appendix C.  Acknowledgements  . . . . . . . . . . . . . . . . . . 29
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 30
   Intellectual Property and Copyright Statements . . . . . . . . . . 31

Mathis & Heffner         Expires April 26, 2006                 [Page 3]

Internet-Draft             Path MTU Discovery               October 2005

1.  Introduction

   This document describes a method for Packetization Layer Path MTU
   Discovery (PLPMTUD) which is an extension to existing Path MTU
   discovery methods as described in RFC 1191 [2] and RFC 1981 [3].  The
   proper MTU is determined by starting with small packets and probing
   with successively larger packets.  The bulk of the algorithm is
   implemented above IP, in the transport layer (e.g.  TCP) or other
   "Packetization Protocol" that is responsible for determining packet

   This document draws heavily RFC 1191 [2] and RFC 1981 [3] for
   terminology, ideas and some of the text.

   This document describes methods to discover the path MTU using
   features of existing protocols.  The methods apply to IPv4 and IPv6,
   and many transport protocols.  They do not require cooperation from
   the lower layers (except that they are consistent about what packet
   sizes are acceptable) or the far node.  Variants in implementations
   will not cause interoperability problems.

   The methods described in this document are carefully designed to
   maximize robustness in the presence of less than ideal
   implementations of other protocols or Internet components.

   For sake of clarity we uniformly prefer TCP and IPv6 terminology.  In
   the terminology section we also present the analogous IPv4 terms and
   concepts for the IPv6 terminology.  In a few situations we describe
   specific details that are different between IPv4 and IPv6.

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   document are to be interpreted as described in RFC 2119 [4].

   This draft is a product of the Path MTU Discovery (pmtud) working
   group of the IETF.  Please send comments and suggestions to  Interim drafts and other useful information will be
   posted at .

1.1.  Revision History

   These are all recent substantive changes, in reverse chronological
   order.  This section will be removed prior to publication as an RFC.
   Note that there are still some missing details that need to be
   resolved.  These are flagged by @@@@.  None of the missing details
   are serious.

Mathis & Heffner         Expires April 26, 2006                 [Page 4]

Internet-Draft             Path MTU Discovery               October 2005

1.1.1.  Changes since version -04, February 2005 (IETF 62)

   General restructuring and rewriting of some sections based on new
   experience.  Relaxed and generalized a lot of over-specified
   language, e.g., the search strategy description.

   Decoupled verification from probing, and relaxed its specification.

   Removed all specified changes to ICMP processing.  We decided this
   was out of scope for this particular document.

   Changed all language to refer to MTU rather than MPS.

1.1.2.  Changes since version -03, October 2004 (IETF 61)

   A number of minor style and grammar edits.

1.1.3.  Changes since version -02, July 19th 2004 (IETF 60)

   Many minor updates throughout the document.

   Added a section describing the interactions between PLPMTUD and
   congestion control.

   Removed a difficult to implement requirement for future data to

   Added "IP Fragmentation" and "Application protocol" as Packetization

   Clarified interactions between TCP SACK and MTU.

   Updated SCTP section to reflect new probing method using "PAD

   Distilled the protocol specific material into separate subsections
   for each protocol.

   Added a section on common requirements and functions for all
   Packetization Layers.  More accurately characterized the
   "bidirectional" (and other) requirements of the PL protocol.  Updated
   the search strategy in this new section.

   Change "ICMP can't fragment" and "packet too big" to uniformly use
   "ICMP PTB message" everywhere.

   Added Stanislav Shalunov's observation that PLPMTUD parallels
   congestion control.

Mathis & Heffner         Expires April 26, 2006                 [Page 5]

Internet-Draft             Path MTU Discovery               October 2005

   Better described the range of interoperability with classical pMTUd
   in the introduction.

   Removed vague language about "not being a protocol" and "excessive

   Slightly redefined flow: the granularity of PLPMTUD within a path.

   Many English NITs and clarifications per Gorry Fairhurst and others.
   Passes strict xml2rfc checking.

   Add a paragraph encouraging interface MTUs that are the optimal for
   the NIC, rather than standard for the media.

   Added a revision history section.

2.  Overview

   This document describes a method for TCP or other packetization
   protocols to dynamically discover the MTU of a path without explicit
   signals from the network.  This method is most efficient when used in
   conjunction with the current ICMP based path MTU discovery mechanism
   as specified in RFC1191 and RFC1981.  When used in such a way, it
   eliminates many robustness problems since it does not depend on the
   delivery ICMP messages.

   These procedures are applicable to TCP and other transport- or
   application-level protocols that are responsible for choosing packet
   boundaries (e.g. segment sizes) and have an acknowledgment structure
   that delivers to the sender accurate and timely indications of which
   packets were lost.

   The general strategy is for the packetization layer to find an
   appropriate path MTU by probing the path with progressively larger
   packets.  If a probe packet is successfully delivered, then the
   effective path MTU is raised to the probe size.

   The isolated loss of a probe packet (with or without an ICMP Packet
   To Big message) is treated as an indication of an MTU limit, and not
   as a congestion indicator.  In this case alone, the packetization
   protocol is permitted to retransmit any missing data without
   adjusting the congestion window.

   If there is a timeout or additional packets are lost during the
   probing process, the probe is considered to be inconclusive (e.g. the
   lost probe does not necessarily indicate that the probe exceeded the
   path MTU).  Furthermore the losses are treated like any other

Mathis & Heffner         Expires April 26, 2006                 [Page 6]

Internet-Draft             Path MTU Discovery               October 2005

   congestion indication: window or rate adjustments are mandatory per
   the relevant congestion control standards [12] Probing can resume
   after a delay which is determined by the nature of the detected

   PLPMTUD uses a searching technique to find the path MTU.  Each
   conclusive probe narrows the MTU search range, either by raising the
   low limit on a successful probe or lowering the high limit on a
   failed probe, until the search range converges toward the true path
   MTU.  For most transport layers, it makes sense to abandon the search
   once the range is narrow enough where the likely gain from picking a
   larger effective path MTU is smaller than the search overhead to find

   The most likely (and least serious) PLPMTUD failure is the link
   experiencing congestion related losses while probing.  In this case
   it is appropriate to retry a probe of the same size as soon as the
   packetization layer has fully adapted to the congestion and recovered
   from the losses.  In other cases, additional losses or timeouts
   indicate problems with the link or packetization layer.  In these
   situations it is desirable to use longer delays depending on the
   severity of the error.

   An optional verification phase can be used to detect some situations
   where raising the MTU raises the packet loss rate.  For example, if a
   link is striped across multiple physical channels with inconsistent
   MTUs, it is possible that a probe will be delivered even if it is too
   large for some of the physical channels.  In such cases raising the
   path MTU to the probe size can cause severe packet loss and abysmal
   performance.  After raising the MTU, the new MTU size can be verified
   by monitoring the loss rate.

   PLPMTUD introduces some flexibility in the implementation of
   classical path MTU discovery, which is subject to protocol failures
   (connection hangs) if ICMP PTB messages are not delivered or
   processed for some reason.  With PLPMTUD, classical path MTU
   discovery can include additional consistency checks (e.g. validating
   additional fields in the transcribed header) without increasing the
   risk of connection hangs due to false failures of the added checks.
   Such changes to classical path MTU discovery are beyond the scope of
   this document.

   In the limiting case, all ICMP PTB messages might be unconditionally
   ignored, and PLPMTUD can be used a the sole method used to discover
   the path MTU.  In this configuration, PLPMTUD parallels congestion
   control.  An end-to-end transport protocol adjusts non-protocol
   properties of the data stream (window size or packet size) while
   using packet losses to deduce the appropriateness of the adjustments.

Mathis & Heffner         Expires April 26, 2006                 [Page 7]

Internet-Draft             Path MTU Discovery               October 2005

   This technique seems to be more philosophically consistent with the
   end-to-end principle of the Internet than relying on ICMP messages
   containing transcribed headers of multiple protocol layers.

   Most of the difficulty in implementing PLPMTUD arises because it
   needs to be implemented in several different places within a single
   node.  In general, each packetization protocol needs to have its own
   implementation of PLPMTUD.  Furthermore, the natural mechanism to
   share path MTU information between concurrent or subsequent
   connections over the same path is a path information cache in the IP
   layer.  The various packetization protocols need to have the means to
   access and update the shared cache in the IP layer.  This memo
   describes PLPMTUD in terms of its primary subsystems without fully
   describing how they are assembled into a complete implementation.

   Section 3 provides a complete glossary of terms.

   Relatively few details of PLPMTUD affect interoperability with other
   standards or Internet protocols.  These details are specified in
   RFC2119 standards language in section 4.  The vast majority of the
   implementation details described in this document are recommendations
   based on experiences with earlier versions of path MTU discovery.
   These recommendations are motivated by a desire to maximize
   robustness of PLPMTUD in the presence of less than ideal network
   conditions as they exist in the field.

   Section 5 describes how to partition PLPMTUD into layers, and how to
   manage the "path information cache" in the IP layer.

   Section 6 describes the general Packetization Layer properties and
   features needed to implement PLPMTUD.

   Section 7 recommends using IPv4 fragmentation in a configuration that
   mimics IPv6 functionality, to minimize future problems migrating to

   Section 8 describes the details of how to use probes to search for
   the path MTU.

   Section 9 describes a programing interface for applications acting as
   packetization layers, and for tools to be able to diagnose path
   problems that interfere with path MTU discovery.

   Section 10 discusses implementation details for specific protocols,
   including TCP.

Mathis & Heffner         Expires April 26, 2006                 [Page 8]

Internet-Draft             Path MTU Discovery               October 2005

3.  Terminology

   We use the following terms in this document:

   IP: Either IPv4 [1] or IPv6 [5].

   Node: A device that implements IP.

   Router: A node that forwards IP packets not explicitly addressed to

   Host: Any node that is not a router.

   Upper layer: A protocol layer immediately above IP.  Examples are
      transport protocols such as TCP and UDP, control protocols such as
      ICMP, routing protocols such as OSPF, and Internet or lower-layer
      protocols being "tunneled" over (i.e., encapsulated in) IP such as
      IPX, AppleTalk, or IP itself.

   Link: A communication facility or medium over which nodes can
      communicate at the link layer, i.e., the layer immediately below
      IP.  Examples are Ethernets (simple or bridged); PPP links; X.25,
      Frame Relay, or ATM networks; and Internet (or higher) layer
      "tunnels", such as tunnels over IPv4 or IPv6.  Occasionally we use
      the slightly more general term "lower layer" for this concept.

   Interface: A node's attachment to a link.

   Address: An IP-layer identifier for an interface or a set of

   Packet: An IP header plus payload.

   MTU: Maximum Transmission Unit, the size in bytes of the largest IP
      packet, including the IP header and payload, that can be
      transmitted on a link or path.  Note that this could more properly
      be called the IP MTU, to be consistent with how other standards
      organizations use the acronym MTU.

   Link MTU: The Maximum Transmission Unit, i.e., maximum IP packet size
      in bytes, that can be conveyed in one piece over a link.  Beware
      that this definition differers from the definition used by other
      standards organizations.

      For IETF documents, link MTU is uniformly defined as the IP MTU
      over the link.  This includes the IP header, but excludes link
      layer headers and other framing which is not part of IP or the IP

Mathis & Heffner         Expires April 26, 2006                 [Page 9]

Internet-Draft             Path MTU Discovery               October 2005

      Be aware that other standards organizations generally define link
      MTU to include the link layer headers.

   Path: The set of links traversed by a packet between a source node
      and a destination node.

   Path MTU, or pMTU: The minimum link MTU of all the links in a path
      between a source node and a destination node.

   Classical path MTU discovery: Process described in RFC 1191 and RFC
      1981, in which nodes rely on ICMP "Packet Too Big" (PTB) messages
      to learn the MTU of a path.

   Packetization layer: The layer of the network stack which segments
      data into packets.

   Effective PMTU: The current estimated value for PMTU used by a
      packetization layer for segmentation.

   PLPMTUD: Packetization Layer Path MTU Discovery, the method described
      in this document, which is an extension to classical PMTU

   PTB (Packet Too Big) message: An ICMP message reporting that an IP
      packet is too large to forward.  This is the IPv6 term that
      corresponds to the IPv4 "ICMP Can't fragment" message.

   Flow: A context in which MTU discovery algorithms can be invoked.
      This is naturally an instance of the packetization protocol, e.g.
      one side of a TCP connection.

   MSS: The TCP Maximum Segment Size [6], the maximum payload size
      available to the TCP layer.  This is typically the path MTU minus
      the size of the IP and TCP headers.

   Probe packet: A packet which is being used to test a path for a
      larger MTU.

   Probe size: The size of a packet being used to probe for a larger

   Probe gap: The payload data that will be lost and need to be
      retransmitted if the probe is not delivered.

Mathis & Heffner         Expires April 26, 2006                [Page 10]

Internet-Draft             Path MTU Discovery               October 2005

   Leading window: Any unacknowledged data in a flow at the time a probe
      is sent.

   Trailing window: Any data in a flow sent after a probe, but before
      the probe is acknowledged.

   Search strategy: The heuristics used to choose successive probe sizes
      to converge on the proper path MTU, as described in section
      Section 8.3.

   Full stop timeout: a timeout where none of the packets transmitted
      after some event are acknowledged by the receiver, including any
      retransmissions.  This is taken as an indication of some failure
      condition in the network, such as a routing change onto a link
      with a smaller MTU.

4.  Requirements

   All Internet nodes SHOULD implement PLPMTUD in order to discover and
   take advantage of the largest MTU supported along the Internet path.

   Links MUST NOT deliver packets that are larger than their MTU.  Links
   that have parametric limitations (e.g.  MTU bounds due to limited
   clock stability) MUST include explicit mechanisms to consistently
   reject packets that might otherwise be nondeterministically

   All hosts SHOULD use IPv4 fragmentation in a mode that mimics IPv6
   functionality.  All fragmentation SHOULD be done on the host, and all
   IPv4 packets, including fragments, SHOULD have the DF bit set such
   that they will not be fragmented (again) in the network.  See
   Section 7.

   The requirements below only apply to those implementations that
   include PLPMTUD.

   To use PLPMTUD a Packetization Layer MUST have a loss reporting
   mechanism that provides the sender with timely and accurate
   indications of which packets were lost in the network.

   Normal congestion control algorithms MUST remain in effect under all
   conditions except when only an isolated probe packet is detected as
   lost.  In this case alone the normal congestion (window or data rate)
   reduction MAY be suppressed.  If any other data loss is detected,
   standard congestion control MUST take place.

   Suppressed congestion control (as above) MUST be rate limited such

Mathis & Heffner         Expires April 26, 2006                [Page 11]

Internet-Draft             Path MTU Discovery               October 2005

   that it occurs less frequently than the worst case loss rate for TCP
   congestion control at a comparable data rate over the same path (i.e.
   less than the "TCP-friendly" loss rate [@@]).  This SHOULD be
   enforced by requiring a minimum headway between a suppressed
   congestion adjustment (due to a failed probe) and the next attempted
   probe, which is equal to one round trip time for each packet
   permitted by the congestion window.  Alternatively this may be
   enforced by not suppressing congestion control if a 2nd probe is lost
   too soon after the 1st lost probe.  This and other issues relating to
   congestion control are discussed in section @@@window.

   Whenever the MTU is raised, the congestion state variables MUST be
   rescaled so as not to raise the window size in bytes (or data rate in
   bytes per seconds).

   Whenever the MTU is reduced (e.g. when processing ICMP PTB messages)
   the congestion state variable SHOULD be rescaled not to raise the
   window size in packets.

   If PLPMTUD updates the MTU for a particular path, all Packetization
   Layer sessions that share the path representation SHOULD be notified
   to make use of the new MTU and make the required congestion

   All implementations MUST include a mechanism to implement diagnostic
   tools that do not rely on the operating systems implementation of
   path MTU discovery.  This specifically requires the ability to send
   packets that are larger than the known MTU for the path, and
   collecting any resultant ICMP error message.  See section 9 for
   further discussion of MTU diagnostics.

5.  Layering

   Packetization Layer Path MTU Discovery is most easily implemented by
   splitting its functions between layers.  The IP layer is the best
   place to keep shared state, collect the ICMP messages, track IP
   header sizes and manage MTU information provided by the link layer
   interfaces.  However, the procedures that PLPMTUD uses for probing,
   verification and scanning for the path MTU are very tightly coupled
   to features of the Packetization Layers such as data recovery and
   congestion control state machines.

   Note that this layering approach is consistent with the advice in the
   current PMTUD specifications [2][3].  Many implementations of
   classical PMTU Discovery are already split along these same layers.

Mathis & Heffner         Expires April 26, 2006                [Page 12]

Internet-Draft             Path MTU Discovery               October 2005

5.1.  Accounting for header sizes

   The way in which PLPMTUD operates across multiple layers requires a
   mechanism for accounting header sizes at all layers between IP and
   the packetization layer (inclusive).  When transmitting non-probe
   packets, it is sufficient for the packetization layer to ensure an
   upper bound on final IP packet size, so as not to exceed the current
   effective path MTU.  All packetization layers participating in
   classical path MTU discovery have this requirement already.  When
   participating in PLPMTUD and transmitting a probe packet, the
   packetization layer must determine that packet's final size including
   IP headers.  This requirement is specific to PLPMTUD, and to satisfy
   it existing implementations may need additional inter-layer

5.2.  Storing PMTU information

   This memo uses the concept of a "flow" to define the scope of the
   path MTU discovery algorithms.  For many implementations, a flow
   would naturally correspond to an instance of each protocol, i.e.,
   each connection or session.  In such implementations the algorithms
   described in this document are performed within each session for each
   protocol.  The observed PMTU can optionally be shared between
   different flows sharing a common path representation.

   Alternatively, PLPMTUD could be implemented such that the complete
   PLPMTUD state is associated with the path representations.  Such an
   implementation could use multiple connections or sessions for each
   probe sequence.  This approach may converge much more quickly in some
   environments such as when an application uses many small connections,
   each of which may be too short to complete the path MTU discovery

   These approaches are not mutually exclusive.  However, due to
   differing constraints on generating probes (section Section 6.2) and
   the MTU searching algorithm (section @@@search), it may not be
   feasible for different packetization layer protocols to share PLPMTUD
   state.  This suggests that it may be possible for some protocols to
   share probing state, but not others.  In this case, the different
   protocols can still share the observed PMTU but they will have
   differing convergence properties.

   The IP layer is the best place to store cached PMTU values and other
   shared state such as MTU values reported by ICMP PTB messages.
   Ideally this shared state should be associated with a specific path
   traversed by packets exchanged between the source and destination
   nodes.  However, in most cases a node will not have enough
   information to completely and accurately identify such a path.

Mathis & Heffner         Expires April 26, 2006                [Page 13]

Internet-Draft             Path MTU Discovery               October 2005

   Rather, a node must associate a PMTU value with some local
   representation of a path.  It is left to the implementation to select
   the local representation of a path.

   An implementation could use the destination address as the local
   representation of a path.  The PMTU value associated with a
   destination would be the minimum PMTU learned across the set of all
   paths in use to that destination.  The set of paths in use to a
   particular destination is expected to be small, in many cases
   consisting of a single path.  This approach will result in the use of
   optimally sized packets on a per-destination basis.  This approach
   integrates nicely with the conceptual model of a host as described in
   [RFC 2461]: a PMTU value could be stored with the corresponding entry
   in the destination cache.  Storing the minimum value is suggested
   since NATs and other forms of middle boxes may exhibit differing
   PMTUs at a single IP address.

   Note that network or subnet numbers are not suitable to use as
   representations of a path, because there is not a general mechanism
   to determine the network mask at the remote host.

   If IPv6 flows are in use, an implementation could use the IPv6 flow
   id [5][9] as the local representation of a path.  Packets sent to a
   particular destination but belonging to different flows may use
   different paths, with the choice of path depending on the flow id.
   This approach will result in the use of optimally sized packets on a
   per-flow basis, providing finer granularity than MTU values
   maintained on a per-destination basis.

   For source routed packets, i.e., packets containing an IPv6 routing
   header, or IPv4 LSRR or SSRR options, the source route may further
   qualify the local representation of a path.  An implementation could
   use source route information in the local representation of a path.

5.3.  Accounting for IPsec

   This document does not take a stance on the placement of IPsec, which
   logically sits between IP and the Packetization Layer.  The PLPMTUD
   implementation can treat IPsec either as part of IP or as part of the
   Packetization Layer, as long as the accounting is consistent within
   the implementation.  If IPsec is treated as part of the IP layer,
   then each security association to a remote node may need to be
   treated as a separate path; i.e., the security association is used to
   represent the path.  If IPsec is treated as part of the packetization
   layer, the IPsec header size must be included in the Packetization
   Layer's header size calculations.  [11]

Mathis & Heffner         Expires April 26, 2006                [Page 14]

Internet-Draft             Path MTU Discovery               October 2005

5.4.  Multicast

   In the case of a multicast destination address, copies of a packet
   may traverse many different paths to reach many different nodes.  The
   local representation of the "path" to a multicast destination must in
   fact represent a potentially large set of paths.

   Minimally, an implementation could maintain a single MTU value to be
   used for all packets originated from the node.  This MTU value would
   be the minimum MTU learned across the set of all paths in use by the
   node.  This approach is likely to result in the use of smaller
   packets than is necessary for many paths.

   If the application using multicast gets complete delivery reports
   (unlikely because this requirement has poor scaling properties),
   PLPMTUD could be implemented in multicast protocols.

6.  Common Packetization Properties

   This section describes general Packetization Layer properties and
   characteristics needed to implement PLPMTUD.  It also describes some
   implementation issues that are common to all Packetization Layers.

6.1.  Mechanism to detect loss

   It is important that the Packetization Layer has a timely and robust
   mechanism for detecting and reporting losses.  PLPMTUD makes MTU
   adjustments on the basis of detected losses.  Any delays or
   inaccuracy in loss notification is likely to result in incorrect MTU
   decisions or slow convergence.

   It is best if Packetization Protocols use fairly explicit loss
   notification such as selective acknowledgments, although implicit
   mechanisms such as TCP Reno style duplicate acknowledgments counting
   are sufficient.  It is important that the mechanism can robustly
   distinguish between the isolated loss of just a probe and other
   combinations of losses.

   Many protocol implementation have complicated mechanisms such as SACK
   scoreboards to distinguish between real losses and temporary missing
   data due to reordering in the network.  In these implementations it
   is desirable to signal losses to PLPMTUD as a side effect of the data
   retransmission.  This approach offers the maximum protection from
   confusing signals due to reordering and other events that might mimic

   PLPMTUD can also be implemented in protocols that rely on timeouts as

Mathis & Heffner         Expires April 26, 2006                [Page 15]

Internet-Draft             Path MTU Discovery               October 2005

   their primary mechanism for loss recovery; however, timeout should be
   used only when there are no other alternatives.

6.2.  Generating probes

   There are several possible ways to alter packetization layers to
   generate probes.  The different techniques incur different overheads
   in three areas: difficulty in generating the probe packet (in terms
   of packetization layer implementation complexity and extra data
   motion) possible additional network capacity consumed by the probes
   and the overhead of recovering from failed probes (both network and
   protocol overheads).

   Some protocols might be extended to allow arbitrary padding with
   dummy data.  This greatly simplifies the implementation because the
   probing can be performed without participation from higher layers and
   if the probe fails, the missing data (the "probe gap") is assured to
   fit within the current MTU when it is retransmitted.  This is
   probably the most appropriate method for protocols that support
   arbitrary length options or multiplexing within the protocol itself.

   Many Packetization Layer protocols can carry pure control messages
   (without any data from higher protocol layers) which can be padded to
   arbitrary lengths.  For example the SCTP HEARTBEAT message can be
   used in this manner (See section 10.2) .  This approach has the
   advantage that nothing needs to be retransmitted if the probe is

   These techniques do not work for TCP, because there is not a separate
   length field or other mechanism to differentiate between padding and
   real payload data.  With TCP the only approach is to send additional
   payload data in an over-sized segment.  There are at least two
   variants of this approach, discussed in section 10.1.

   In a few cases there may no reasonable mechanisms to generate probes
   within the Packetization Layer protocol itself.  As a last resort it
   may be possible to rely an an adjunct protocol, such as ICMP ECHO
   (aka "ping"), to send probe packets.  See section 10.3 for further
   discussion of this approach.

7.  Host Fragmentation

   Packetization layers are encouraged to avoid sending messages that
   will require fragmentation.  (For the case against fragmentation, see
   [14], [15]).  However, entirely preventing fragmentation is not
   always possible.  Some packetization layers, such as a UDP
   application outside the kernel, may be unable to change the size of

Mathis & Heffner         Expires April 26, 2006                [Page 16]

Internet-Draft             Path MTU Discovery               October 2005

   messages it sends, resulting in datagram sizes that exceed the path

   IPv4 permitted such applications to send packets without the DF bit
   set.  Oversized packets without the DF bit set would be fragmented in
   the network or sending host when they encountered a link with a MTU
   smaller than the packet.  In some case, packets could be fragmented
   more than once if there were cascaded links with progressively
   smaller MTUs.  This approach is not recommended.

   It is recommended that IPv4 implementations use a strategy that
   mimics IPv6 functionality.  When an application sends datagrams that
   are larger than the known path MTU they should be fragmented to the
   path MTU in the host IP layer even if they are smaller than the link
   MTU of the first network hop directly attached to the host.  The DF
   bit should be set on the fragments, so they will not be fragmented
   again in the network.

   This technique will minimize future surprises as the Internet
   migrates to IPv6.  Otherwise, the potential exists for widely
   deployed applications or services relying on IPv4 fragmentation in a
   way that cannot be implemented in IPv6.  At least one major operating
   system already uses this strategy.

   The ability to selectively transmit packets larger than the current
   effective path MTU (but smaller than the link MTU) is required, to be
   able to send probes generated by packetization layers participating
   in PLPMTUD, and to facilitate diagnostic utilities.

   Note that IP fragmentation divides data into packets, so it is
   minimally a Packetization Layer.  However, it does not have a
   mechanism to detect lost packets, so it can not support a native
   implementation of PLPMTUD.  Fragmentation-based PLPMTUD requires an
   adjunct protocol as described in section 10.3.

8.  The Probing Method

   This section describes the details of the MTU probing method,
   including how to send probes and process error indications necessary
   to search for the path MTU.

8.1.  Packet size ranges

   A packetization layer implementing PLPMTUD should keep three pieces
   of state:

Mathis & Heffner         Expires April 26, 2006                [Page 17]

Internet-Draft             Path MTU Discovery               October 2005

   search_low: The smallest available probe size, minus one.
   search_high: The greatest available probe size.
   eff_pmtu: The effective pmtu for this flow.

               search_low          eff_pmtu         search_high
                   |                   |                  |
               non-probe size range
                               probe size range

   Figure 1

   When transmitting probes, the packetization layer will select the
   probe size from within the range "(search_low, search_high]".  When
   transmitting non-probes, it may use sizes less than or equal to

   The eff_pmtu must be in the range "[search_low, search_high]".  Note
   that when probing upward eff_pmtu will equal search_low, but may
   differ due to initial values, or ICMP PTB messages.

8.2.  Selecting initial values

   The initial value for search_high should be the largest possible
   packet supported by the flow.  This may be limited by the local
   interface MTU, by a protocol mechanism such as the TCP MSS option, or
   an intrinsic limit such as the protocol length field.

   It is recommended that search_low be initially set to a value likely
   to work over a large range of links.  Given today's technologies, a
   value of 512 bytes is likely to work.  For IPv6 flows, a value of
   1280 is appropriate.  The initial value for search_low should be

   Properly functioning Path MTU discovery is critical to the robust and
   efficient operation of the Internet.  Any major change (as described
   in this document) has the potential to be very disruptive if it
   contains any errors or oversights.  The selection of initial values
   determines to what extent a PLPMTUD implementation's behavior differs
   from classical PMTUD in cases where MTU discovery is not needed, or
   where classical PMTUD is sufficient.

   It may be desirable to configure hosts in such a way that PLPMTUD
   only has an effect in cases where classical PMTUD fails.  Setting
   eff_pmtu equal to search_high and relying on black hole detection has
   this effect.  Using initial values of search_low = eff_pmtu =
   search_high has the effect of effectively disabling PLPMTUD and

Mathis & Heffner         Expires April 26, 2006                [Page 18]

Internet-Draft             Path MTU Discovery               October 2005

   relying only on classical PMTUD.

   In some cases where it is known that classical PMTUD is likely to
   fail, using a conservatively small initial eff_pmtu may produce
   better results.  Using a small initial eff_pmtu may have na impact on
   protocol dynamics in all cases, but can result in much better
   performance by avoiding the costly timeouts required for black hole

   As appropriate initial values for PLPMTUD state variables may vary
   not only per host but per path, configuration options for these
   values in the route cache is desirable.

8.3.  Selecting probe size

   The probe may have a size anywhere in the "probe size range"
   described above.  However, a number of factors affect the selection
   of an appropriate size.  A simple strategy might be to do a binary
   search halving the probe size range with each probe.  However, for
   some protocols, data in a lost probe may require retransmission,
   making a failed probe more expensive than a successful probe.  For
   such protocols, a strategy using smaller probe sizes and "probing up"
   may behave better.  For many protocols, both at and above the
   packetization layer, the benefit of increasing MTU sizes may follow a
   step function such that it is not advantageous to probe within
   certain regions at all.

   As an optimization, it may be appropriate to probe at certain common
   or expected MTU sizes; for example, 1500 bytes for standard Ethernet,
   or 1500 bytes minus header sizes for tunnel protocols.

   Some protocols may not even "choose" probe sizes.  For protocols
   which have certain natural data block sizes, an effective strategy
   could be to simply treat blocks whose size falls in the probe size
   range as a probe.

   Each packetization layer must determine when probing is considered
   converged; that is, when the probe size range is considered small
   enough that further probing is no longer worth its cost.  When it is
   determined that searching has converged, a timer should be set for 5
   minutes @@why.  When the timer expires, search_high should be reset
   to its initial value (described above) so that probing can resume.
   This is so that if the path changes, and in increased path MTU is
   available, then the flow will eventually be able to take advantage of
   it to send larger packets.

Mathis & Heffner         Expires April 26, 2006                [Page 19]

Internet-Draft             Path MTU Discovery               October 2005

8.4.  Probing preconditions

   Before sending a probe, the flow must at least meet the following
   o  The flow has no outstanding probes or losses.
   o  If the last probe failed or was inconclusive, then the probe
      timeout has expired (see Section 8.6.2).
   o  The available window is greater than the probe size.
   o  For a protocol which uses in-band data for probing, enough data is
      available to send the probe.

   In order to allow loss detection mechanisms to be effective, some
   protocols may require a probe plus a number of non-probe packet's
   worth of available data and window space.

   When not enough data is available to probe, the protocol may wish to
   delay sending non-probes in order to accumulate enough data to send a

8.5.  Conducting a probe

   Once a probe size in the appropriate range has been selected, and the
   above preconditions have been met, the packetization layer may
   conduct a probe.  To do so, it creates a probe packet such that its
   size, including the outermost IP headers, is equal the probe size.
   After sending the probe it awaits response, which may take the
   following results:
   Success: The probe is acknowledged as having been received by the
      remote host.

   Failure: A protocol mechanism indicates that the probe was lost, but
      no packets in the leading or trailing window were lost.

   Timeout failure: A protocol mechanism indicates that the probe was
      lost, and no packets in the leading window were lost, but is
      unable to determine if any packets in the trailing window were
      lost.  For example, loss is detected by a timeout, and go-back-n
      retransmission is used.

   Inconclusive: The probe was lost in addition to other packets in the
      leading or trailing windows.

8.6.  Response to probe results

   When a probe has completed, the result should be processed as
   follows, categorized by the probe's result type.

Mathis & Heffner         Expires April 26, 2006                [Page 20]

Internet-Draft             Path MTU Discovery               October 2005

8.6.1.  Probe success

   When the probe is delivered, this is an indication that the path MTU
   is at least as large as the probe size.  The packetization layer
   should set search_low to the probe size, eff_pmtu to "max(eff_pmtu,
   probe size)".

   Note that if a flow's packets are routed via multiple paths, or over
   a path with a non-deterministic MTU, delivery of a single probe
   packet does not indicate that all packets of that size will be
   delivered.  To be robust in such a case, the packetization layer
   should conduct MTU verification as described in section @@cite.

8.6.2.  Probe failure

   When only the probe is lost, this is treated as an indication that
   the path MTU is smaller than the probe size.  In this case alone, the
   loss should not be interpreted as congestion signal.

   In the absence of other indications, the packetization layer should
   set search_high to the probe size minus one, and eff_pmtu to
   "min(eff_pmtu, probe size)".

   If an ICMP PTB message is received matching the probe packet, then
   search_high and eff_pmtu may be set from the MTU value indicated in
   the message.  Note that the ICMP message may be received either
   before or after the protocol loss indication.

   A probe failure event is the one situation under which the
   packetization layer is permitted not to treat loss as a congestion
   signal.  Because there is some small risk that suppressing congestion
   control might have unanticipated consequences (even for one isolated
   loss), it is required that probe failure events be less frequent than
   the normal period for losses under standard congestion control.
   Specifically after a probe failure event and suppressed congestion
   control, PLPMTUD may not probe again until an interval which is
   comparable to the expected interval between congestion control
   events.  This is required in section 4 and discussed further in
   section @@@window.  The simplest estimate of the interval to the next
   congestion event is the same number of round trips as the current
   congestion window in packets.

8.6.3.  Probe timeout failure

   If the loss was detected with a timeout and repaired with go-back-n
   retransmission, then congestion window reduction will be necessary.
   The relatively high price of a failed probe in this case may merit a
   longer timeout.  A timeout value of five times @@why? the non-timeout

Mathis & Heffner         Expires April 26, 2006                [Page 21]

Internet-Draft             Path MTU Discovery               October 2005

   failure case is recommended.

8.6.4.  Probe inconclusive

   The presence of other losses near the loss of the probe may indicate
   that the probe was lost due to congestion rather than because of an
   MTU limitation.  In this case it is appropriate to update no state,
   and simply probe again when the probing preconditions are met; i.e.,
   when no recent losses have been observed.  At this point, it is
   particularly appropriate to re-probe since the flow's congestion
   window will be at its lowest point, minimizing the probability of
   congestive losses.

8.7.  Full stop timeout

   Under all conditions a full stop timeout (also known as a "persistent
   timeout" in other documents) should be taken as an indication of some
   significantly disruptive event in the network, such as a router
   failure or a routing change to a path with a smaller MTU.  For TCP,
   this occurs when the R1 timeout threshold described by [8] expires.

   If there is a full stop timeout and there was not an ICMP message
   indicating a reason (PTB, Net unreachable, etc, or the ICMP messages
   was ignored for some reason), the suggested first recovery action is
   to treat this as a detected black hole as described in [10].

   The response to a detected black hole should be to set search_low to
   its initial value, and set eff_pmtu to search_low.  Upon further
   successive timeouts, search_low and eff_pmtu should be halved, with a
   lower bound of 68 bytes for IPv4 and 1280 bytes for IPv6.

8.8.  MTU verification

   It is possible for a flow to simultaneously traverse multiple paths,
   but it will only be able to keep a single path representation for the
   flow.  If in such a case the paths have different MTUs, storing the
   minimum MTU of all paths in the flow's path representation will
   result in correct, though sub-optimal behavior.  If ICMP PTB messages
   are delivered, then classical PMTUD will work correctly in this

   However, if ICMP is not delivered, PLPMTUD will be relied upon and
   may fail because its requirement that links MUST NOT deliver packets
   larger than their MTU is effectively broken.  A probe with a size
   greater than the minimum but smaller than the maximum of the path
   MTUs may be successful.  However, upon raising the flow's effective
   PMTU, the loss rate may significantly increase.  The flow may still
   make progress, but the resultant loss rate may be unacceptable.  For

Mathis & Heffner         Expires April 26, 2006                [Page 22]

Internet-Draft             Path MTU Discovery               October 2005

   example, when using two-way round-robin striping, 50% of full-sized
   packets would be lost.

   Striping in this manner is often operationally undesirable (for
   example, due to packet reordering), and is usually avoided by hashing
   flows to a single path.  However, to increase robustness an
   implementation should implement some form of MTU verification, such
   that if increasing eff_pmtu results in a sharp increase in loss rate,
   it will fall back to using a lower MTU.

   A recommended strategy would be that when using a new value for
   eff_pmtu, to save the old value.  If loss rate rises above a certain
   threshold for a period of time (for example, loss rate is higher than
   10% over multiple RTO intervals), then the new MTU is considered
   incorrect.  The saved value of eff_pmtu should be restored, and
   search_high reduced in the same manner as in a probe failure.

9.  Diagnostic Interface

   All implementations MUST include facilities for MTU discovery
   diagnostic tools that implement PLPMTUD or other MTU discovery
   algorithms in user mode without help or interference by the PMTUD
   algorithm present in the operating system.  This requires a mechanism
   where a diagnostic application can send packets that are larger than
   the operating system's notion of the current path MTU and for the
   diagnostic application to collect any resulting ICMP PTB messages or
   other ICMP messages.  For IPv4, the diagnostic application must be
   able to set the DF bit.

   At this time nearly all operating systems support two modes for
   sending UDP datagrams: one which silently fragments packets that are
   too large, and another that rejects packets that are too large.
   Neither of these modes are suitable for efficiently diagnosing
   problems with MTU discovery, such as routers that return ICMP PTB
   messages containing incorrect size information.

10.  Specific Packetization Layers

   This section discusses specific implementation details for different
   protocols that can be used as Packetization Layer protocols.  All
   Packetization Layer protocols must consider all of the issues
   discussed in section Section 6.  For most protocols it is self
   evident how to address many of these issues.  It is hoped that the
   protocols described here will be sufficient illustration for
   implementors to adapt other protocols.

Mathis & Heffner         Expires April 26, 2006                [Page 23]

Internet-Draft             Path MTU Discovery               October 2005

10.1.  Probing method using TCP

   TCP has no mechanism that could be used to distinguish between real
   application data and some other form of padding that might be used to
   fill out probe packets.  Therefore, TCP must generate probes by
   sending oversized segments that are carrying real data from upper
   layers.  There are two approaches that TCP might use to minimize
   overhead associated with the probing sequence.

   A TCP implementation of PLPMTUD can elect to send subsequent segments
   overlapping the probe as though the probe segment was not oversized.
   This has the advantage that TCP only needs to retransmit one segment
   at the current MTU to recover from failed probes.  However the
   duplicate data in the probe does consume network resources and will
   cause duplicate acknowledgments.  It is important that these extra
   duplicate acknowledgments not trigger Fast Retransmit.  This can be
   guaranteed by limiting the largest probe segment size to twice the
   current segment size (causing at most 1 duplicate acknowledgment) or
   three times the current segment size (causing at most 2 duplicate

   The other approach is to send non-overlapping segments following the
   probe.  Although this is cleaner from a protocol architecture
   standpoint it clashes with many of the optimizations used to improve
   the efficiency of data motion within many operating systems.  In
   particular many implementations divide the data into segments and
   pre-compute checksums as the data is copied out of application
   buffers.  In these implementations it can be relatively expensive to
   adjust segment boundaries after the data is already queued.

10.2.  Probing method using SCTP

   In the SCTP protocol [7][13] the application writes messages to SCTP
   and SCTP "chunkifies" them into smaller pieces suitable for
   transmission through the network.  Once a message has been
   chunkified, they are assigned Transmission Sequence Numbers (TSNs).
   Once some TSNs have been transmitted SCTP can not change the chunk
   sizes.  SCTP multi-path support normally requires SCTP to chunkify
   its messages to fit the smallest PMTU of all paths.  Although not
   required, implementations may bundle multiple data chunks together to
   make larger IP packets to send on paths with a larger PMTU.  Note
   that SCTP must independently probe the PMTU on each path to the peer.

   The recommended method for generating probes is to add a chunk
   consisting only of padding to an SCTP message.  There are two methods
   to implement this padding.

   In method 1, the message is padded with an SCTP heart beat (HB), of

Mathis & Heffner         Expires April 26, 2006                [Page 24]

Internet-Draft             Path MTU Discovery               October 2005

   the necessary size to construct an IP packet the desired probe size.
   The peer SCTP implementation will acknowledge a successful probe
   without delay by returning the same Heartbeat as a HEARTBEAT-ACK.
   This method is fully compatible with current SCTP standards and
   implementations, but is exposed to MTU limitation on the return path,
   which might cause the HEARTBEAT-ACK to be lost.

   In method 2, a new "PAD" chunk type would have to be defined.  This
   chunk would be silently discarded by the peer.  The PAD chunk could
   be attached to another message (either a minimum length HB or other
   application data which will be acknowledged by the peer) to build a
   probe packet.  The default action for an unknown chunk type in the
   range 128 to 190, (high bits = 10 ) is to "Skip this chunk and
   continue processing" [RFC2960] - exactly the required behavior for a
   PAD chunk.  Any currently unused type in this range will work for a
   PAD chunk type.  This method is fully compatible with all current
   SCTP implementations, but requires adding a new type to the current
   standards.  It has the advantage that restrictions due to the return
   path MTU are not applied to the forward path.

10.3.  Probing method using IP fragmentation

   As mentioned in section 7, datagram protocols (such as UDP) might
   rely on IP fragmentation as a packetization layer.  However,
   implementing PLPMTUD with IP fragmentation is problematic because the
   IP layer has no mechanism to determine if the packets are ultimately
   delivered properly to the far node, without participation by the

   To support IP fragmentation as a packetization layer under an
   unmodified application, we propose the use of an adjunct MTU
   measurement protocol (ICMP ECHO) and a separate path MTU discovery
   daemon (described here) to perform PLPMTUD and update the stored path
   MTU information.

   For IP fragmentation the initial MTU should be selected as described
   in section 8.2, except with a separate global control for the default
   initial MTU for connectionless protocols.  Since connectionless
   protocols may not keep enough state to effectively diagnose MTU black
   holes, it would be more robust to error on the side of using too
   small of an initial MTU (e.g. 1kBytes or less) prior to initiating
   probing of the path to measure the MTU.

   Since many protocols that rely on IP fragmentation are
   connectionless, there is an additional problem with the path
   information cache: there are no events corresponding to connection
   establishment and tear-down to use to manage the cache itself.  If
   there is no entry in the path information cache for a particular

Mathis & Heffner         Expires April 26, 2006                [Page 25]

Internet-Draft             Path MTU Discovery               October 2005

   packet being transmitted, it uses an immutable cache entry for the
   "default path", which has a MTU that is fixed at the initial value.
   A new path cache entry is not created until there is an attempt to
   set the MTU.

   The path MTU discovery daemon should be triggered as a side effect of
   IP fragmentation.  Once the number of fragmented datagrams via any
   particular path reaches some configurable threshold (say 5
   datagrams), the daemon can start probing the path with ICMP ECHO
   packets.  These probes must use the diagnostic interface described in
   section 9 and have DF set.  The daemon can implement all of the
   PLPMTUD probe sequence and search strategy, collect all of the ICMP
   responses (ECHO REPLY, ICMP PTB, etc) and only the saved PTB in the
   path information cache in the IP layer.

   Alternatively, most of the PLPMTUD state machinery can be implemented
   within the path information cache in the IP layer, which can
   specifically invoke the path MTU discovery daemon to perform
   specified measurements on specific paths and report the results back
   to the IP layer.

   Using ICMP ECHO to measure the MTU has a number of potential
   robustness problems.  Note that the most likely failures are due to
   losses unrelated to MTU (e.g. nodes that discriminate on the basis of
   protocol type).  These non-MTU-related losses can prevent PLPMTUD
   from raising the MTU, forcing the packetization protocol to use a
   smaller MTU than necessary.  Since these failures are not likely to
   cause interoperability problems they are relatively benign.

   However there does exist other more serious failure modes, such as
   layer 3 or 4 routers choosing different paths for different protocol
   types or sessions.  In such environments, adjunct protocols may
   experience different MTUs than the primary protocol.  If the adjunct
   protocol has a larger MTU than the primary protocol, PLPMTUD will
   select a non-functional MTU.  This does not seem to be a likely

10.4.  Probing method using applications

   The disadvantages of probing with ICMP ECHO can be overcome by
   implementing the path MTU discovery daemon within the application
   itself, using the application's own protocol.

   The application must have some suitable method for generating probes.
   The ideal situation is a lightweight echo function, that confirms
   message delivery, plus a mechanism for padding the messages out to
   the desired MTU, such that the padding is not echoed.  This
   combination (akin to the SCTP HB plus PAD) is preferred because you

Mathis & Heffner         Expires April 26, 2006                [Page 26]

Internet-Draft             Path MTU Discovery               October 2005

   can send large probes that cause small acknowledgments.  For
   protocols that can not implement these messages directly there are
   often alternate methods for generating probes.  For example, the
   protocol may have a variable length echo (that measures both the
   forward and return path) or if there is no echo function, there may
   be a way to add padding to regular messages carrying real application
   data.  There may also be other ways to generate probes.  As a last
   resort, it may be feasible to extend the protocol with new message
   types to support MTU discovery.

   Probing within an application introduces one new issue: many
   applications do not currently concern themselves with MTU and rely on
   IP fragmentation to deliver datagrams that just happen to be larger
   than the path MTU.  PLPMTUD requires that the protocol be able to
   send probes that are larger than the IP layer's current notion of the
   path MTU, but are marked not to be fragmented.  This requires an
   alternate method for sending these datagrams.

   As with ICMP MTU probing, there is considerable flexibility in how
   the PLPMTUD algorithms can be divided between the Application and the
   path information cache.

   Some applications send large datagrams no matter what the link size,
   and rely on IP fragmentation to deliver the datagrams.  It has been
   known for a long time that this has some undesirable consequences
   [@@harm1].  Recently it has come to light that IPv4 fragmentation is
   not sufficiently robust for general use in today's Internet.  The 16-
   bit IP identification field is not large enough to prevent frequent
   misassociated IP fragments and the TCP and UDP checksums are
   insufficient to prevent the resulting corrupted data from being
   delivered to higher protocol layers. [@@harm2]

   None the less, there are a number of higher layer protocols, such as
   NFS [@@NFS] which use IP fragmentation as a mechanism to reduce CPU
   load.  NFS typically sends fragmented 8k Byte datagram's over all
   link types, no matter what the link MTU.  The other common case, in
   which the application wants to use the largest possible datagram that
   fits within the MTU is most easily treated as a special case of the
   fragmenting case.

11.  References

11.1.  Normative references

   [1]  Postel, J., "Internet Protocol", STD 5, RFC 791, September 1981.

   [2]  Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191,

Mathis & Heffner         Expires April 26, 2006                [Page 27]

Internet-Draft             Path MTU Discovery               October 2005

        November 1990.

   [3]  McCann, J., Deering, S., and J. Mogul, "Path MTU Discovery for
        IP version 6", RFC 1981, August 1996.

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

   [5]  Deering, S. and R. Hinden, "Internet Protocol, Version 6 (IPv6)
        Specification", RFC 2460, December 1998.

   [6]  Postel, J., "Transmission Control Protocol", STD 7, RFC 793,
        September 1981.

   [7]  Stewart, R., Xie, Q., Morneault, K., Sharp, C., Schwarzbauer,
        H., Taylor, T., Rytina, I., Kalla, M., Zhang, L., and V. Paxson,
        "Stream Control Transmission Protocol", RFC 2960, October 2000.

   [8]  Braden, R., "Requirements for Internet Hosts - Communication
        Layers", STD 3, RFC 1122, October 1989.

11.2.  Informative references

   [9]   Partridge, C., "Using the Flow Label Field in IPv6", RFC 1809,
         June 1995.

   [10]  Lahey, K., "TCP Problems with Path MTU Discovery", RFC 2923,
         September 2000.

   [11]  Kent, S. and R. Atkinson, "Security Architecture for the
         Internet Protocol", RFC 2401, November 1998.

   [12]  Floyd, S., "Congestion Control Principles", BCP 41, RFC 2914,
         September 2000.

   [13]  Stewart, R., "Stream Control Transmission Protocol (SCTP)
         Implementors Guide", draft-ietf-tsvwg-sctpimpguide-10 (work in
         progress), December 2003.

   [14]  Kent, C. and J. Mogul, "Fragmentation considered harmful",
         Proc. SIGCOMM '87 vol. 17, No. 5, October 1987.

   [15]  Mathis, M., Heffner, J., and B. Chandler, "Fragmentation
         Considered Very Harmful", draft-mathis-frag-harmful-00 (work in
         progress), July 2004.

Mathis & Heffner         Expires April 26, 2006                [Page 28]

Internet-Draft             Path MTU Discovery               October 2005

Appendix A.  Security Considerations

   Under all conditions the PLPMTUD procedure described in this document
   is at least as secure as the current standard path MTU discovery
   procedures described in RFC 1191 [2] and RFC 1981 [3].

   Since this algorithm is designed for robust operation without any
   ICMP (or other messages from the network), PLPMTUD could be
   configured to ignore all ICMP messages (globally or on a per
   application basis).  In this configuration, it cannot be attacked
   unless the attacker can identify and selectively cause probe packets
   to be lost.

Appendix B.  IANA Considerations


Appendix C.  Acknowledgements

   Many ideas and even some of the text come directly from RFC1191 and

   Many people made significant contributions to this document,
   including: Randall Stewart for SCTP text, Michael Richardson for
   material from an earlier ID on tunnels that ignore DF, Stanislav
   Shalunov for the idea that pure PLPMTUD parallels congestion control,
   and Matt Zekauskas for maintaining focus during the meetings.  Thanks
   to the early implementors: Kevin Lahey, John Heffner and Rao Shoaib
   who provided concrete feedback on weaknesses in earlier drafts.
   Thanks also to all of the people who made constructive comments in
   the working group meetings and on the mailing list.  I am sure I have
   missed many deserving people.

   Matt Mathis and John Heffner are supported in this work by a grant
   from Cisco Systems, Inc.

Mathis & Heffner         Expires April 26, 2006                [Page 29]

Internet-Draft             Path MTU Discovery               October 2005

Authors' Addresses

   Matt Mathis
   Pittsburgh Supercomputing Center
   4400 Fifth Avenue
   Pittsburgh, PA  15213

   Phone: 412-268-3319

   John W. Heffner
   Pittsburgh Supercomputing Center
   4400 Fifth Avenue
   Pittsburgh, PA  15213

   Phone: 412-268-2329

Mathis & Heffner         Expires April 26, 2006                [Page 30]

Internet-Draft             Path MTU Discovery               October 2005

Intellectual Property Statement

   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights 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; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in BCP 78 and BCP 79.

   Copies of IPR disclosures made to the IETF Secretariat 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 implementers or users of this
   specification can be obtained from the IETF on-line IPR repository at

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

Disclaimer of Validity

   This document and the information contained herein are provided on an

Copyright Statement

   Copyright (C) The Internet Society (2005).  This document is subject
   to the rights, licenses and restrictions contained in BCP 78, and
   except as set forth therein, the authors retain all their rights.


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

Mathis & Heffner         Expires April 26, 2006                [Page 31]