RFC 4951






Network Working Group                                       V. Jain, Ed.
Request for Comments: 4951                           Riverstone Networks
Category: Standards Track                                    August 2007


Fail Over Extensions for Layer 2 Tunneling Protocol (L2TP) "failover"

Status of This Memo



   This document specifies an Internet standards track protocol for the
   Internet community, and requests discussion and suggestions for
   improvements.  Please refer to the current edition of the "Internet
   Official Protocol Standards" (STD 1) for the standardization state
   and status of this protocol.  Distribution of this memo is unlimited.

Copyright Notice



   Copyright (C) The IETF Trust (2007).

Abstract



   Layer 2 Tunneling Protocol (L2TP) is a connection-oriented protocol
   that has a shared state between active endpoints.  Some of this
   shared state is vital for operation, but may be volatile in nature,
   such as packet sequence numbers used on the L2TP Control Connection.
   When failure of one side of a control connection occurs, a new
   control connection is created and associated with the old connection
   by exchanging information about the old connection.  Such a mechanism
   is not intended as a replacement for an active fail over with some
   mirrored connection states, but as an aid for those parameters that
   are particularly difficult to have immediately available.  Protocol
   extensions to L2TP defined in this document are intended to
   facilitate state recovery, providing additional resiliency in an L2TP
   network, and improving a remote system's layer 2 connectivity.

















Jain, et al.                Standards Track                     [Page 1]

RFC 4951                        FAILOVER                     August 2007


Table of Contents



   1. Introduction ....................................................3
      1.1. Terminology ................................................4
      1.2. Abbreviations ..............................................5
      1.3. Specification of Requirements ..............................5
   2. Overview ........................................................5
   3. Failover Protocol ...............................................7
      3.1. Failover Capability Negotiation ............................7
      3.2. Failover Recovery Procedure ................................7
           3.2.1. Recovery Tunnel Establishment .......................8
           3.2.2. Control Channel Reset ..............................10
           3.2.3. Data Channel Reset .................................10
      3.3. Session State Synchronization .............................11
   4. New Control Messages ...........................................12
      4.1. Failover Session Query ....................................13
      4.2. Failover Session Response .................................13
   5. New Attribute Value Pairs ......................................14
      5.1. Failover Capability AVP ...................................14
      5.2. Tunnel Recovery AVP .......................................15
      5.3. Suggested Control Sequence AVP ............................16
      5.4. Failover Session State AVP ................................17
   6. Configuration Parameters .......................................18
   7. IANA Considerations ............................................19
   8. Security Considerations ........................................19
   9. Acknowledgements ...............................................19
   10. Contributors ..................................................20
   11. References ....................................................20
      11.1. Normative References .....................................20
      11.2. Informative References ...................................20
   Appendix A ........................................................21
   Appendix B ........................................................23
   Appendix C ........................................................24


















Jain, et al.                Standards Track                     [Page 2]

RFC 4951                        FAILOVER                     August 2007


1.  Introduction



   The goal of this document is to aid the overall resiliency of an L2TP
   endpoint by introducing extensions to RFC 2661 [L2TPv2] and RFC 3931
   [L2TPv3] that will minimize the recovery time of the L2TP layer after
   a failover, while minimizing the impact on its performance.
   Therefore, it is assumed that the endpoint's overall architecture is
   also supportive in the resiliency effort.

   To ensure proper operation of an L2TP endpoint after a failover, the
   associated information of the control connection and sessions between
   them must be correct and consistent.  This includes both the
   configured and dynamic information.  The configured information is
   assumed to be correct and consistent after a failover, otherwise the
   tunnels and sessions would not have been setup in the first place.

   The dynamic information, which is also referred to as stateful
   information, changes with the processing of the tunnel's control and
   data packets.  Currently, the only such information that is essential
   to the tunnel's operation is its sequence numbers.  For the tunnel
   control channel, the inconsistencies in its sequence numbers can
   result in the termination of the entire tunnel.  For tunnel sessions,
   the inconsistency in its sequence numbers, when used, can cause
   significant data loss, which gives the perception of a "service loss"
   to the end user.

   Thus, an optimal resilient architecture that aims to minimize
   "service loss" after a failover, must make provisions for the
   tunnel's essential stateful information, i.e., its sequence numbers.
   Currently, there are two options available: the first option is to
   ensure that the backup endpoint is completely synchronized with the
   active endpoint, with respect to the control and data sessions
   sequence numbers.  The other option is to reestablish all the tunnels
   and their sessions after a failover.  The drawback of the first
   option is that it adds significant performance and complexity impact
   to the endpoint's architecture, especially as tunnel and session
   aggregation increases.  The drawback of the second option is that it
   increases the "service loss" time, especially as the architecture
   scales.

   To alleviate the above-mentioned drawbacks of the current options,
   this document introduces a mechanism to bring the dynamic stateful
   information of a tunnel to a correct and consistent state after a
   failure.  The proposed mechanism defines the recovery of tunnels and
   sessions that were in an established state prior to the failure.






Jain, et al.                Standards Track                     [Page 3]

RFC 4951                        FAILOVER                     August 2007


1.1.  Terminology



   Endpoint: L2TP control connection endpoint, i.e., either LAC or LNS
   (also known as LCCE in [L2TPv3]).

   Active Endpoint: An endpoint that is currently providing service.

   Backup Endpoint: A redundant endpoint standing by for the active
   endpoint that has its database of active tunnels and sessions in sync
   with its active endpoint.

   Failed Endpoint: The endpoint that was the active endpoint at the
   time of the failure.

   Recovery Endpoint: The endpoint that initiates the failover protocol
   to recover from the failure of an active endpoint.

   Remote Endpoint: The endpoint that peers with active endpoint before
   failure and with recovery endpoint after failure.

   Failover: The action of a backup endpoint taking over the service of
   an active endpoint.  This could be due to administrative action or
   failure of the active endpoint.

   Old Tunnel: A control connection that existed before failure and is
   subjected to recovery upon failover.

   Recovery Tunnel: A new control connection established only to recover
   an old tunnel.

   Recovered Tunnel: After an old tunnel's control connection and
   sessions are restored using the mechanism described in this document,
   it is referred to as a Recovered Tunnel.

   Control Channel Failure: Failure of the component responsible for
   establishing/maintaining tunnels and sessions at an endpoint.

   Data Channel Failure: Failure of the component responsible for
   forwarding the L2TP encapsulated data.












Jain, et al.                Standards Track                     [Page 4]

RFC 4951                        FAILOVER                     August 2007


1.2.  Abbreviations



   LAC      L2TP Access Concentrator
   LNS      L2TP Network Server
   LCCE     L2TP Control Connection Endpoint
   AVP      Attribute Value Pair
   SCCRQ    Start-Control-Connection-Request
   SCCRP    Start-Control-Connection-Reply
   ZLB      Zero Length Body Message

1.3.  Specification of Requirements



   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 [RFC2119].

2.  Overview



   The following diagram depicts the redundancy architecture and
   pertaining entities used to describe the failover protocol:

                                           +--------------+
                                           | L2TP active  |
   +----------+                        ----| endpoint (A) |
   |   L2TP   |                       /    +--------------+
   | endpoint |----------------------/
   |    (R)   |                      \     +--------------+
   +----------+                       \    | L2TP backup  |
                                       ----| endpoint (B) |
                                           +--------------+

   Active and backup endpoints may reside on the same device, however,
   they are not required to be that way.  On other hand, some devices
   may not have a standby module altogether, in which case the failed
   endpoint, after reset, can become the recovery endpoint to recover
   from its prior failure.

   Therefore, in the above diagram, upon A's (active endpoint's)
   failure:

      - Endpoint A would be called the failed endpoint.

      - If B is present, then it would become the recovery endpoint and
        also an active endpoint.

      - If B is not present, then A could become the recovery endpoint
        after it restarts, provided it saved the information about
        active tunnels/sessions in some persistent storage.



Jain, et al.                Standards Track                     [Page 5]

RFC 4951                        FAILOVER                     August 2007


      - R does not initiate the failover protocol; rather it waits for a
        failure indication from recovery endpoint.

   This document assumes that the actual detection of a failure in the
   backup endpoint is done in an implementation-specific way.  It also
   assumes that failure detection by the backup endpoint is faster than
   the L2TP control channel timeout between the active and remote
   endpoints.  Typically, active and backup endpoints reside on the same
   physical device, allowing fast and reliable failure detection without
   the need to communicate between these endpoints over the network.

   Similarly, an active endpoint that acts also as a backup endpoint can
   easily start the recovery once it has rebooted.  However, when the
   active and backup endpoints reside in separate devices, there is a
   need for communication between them in order to detect failures.  As
   a solution for such situations, additional failure detection
   protocols, e.g., [BFD-MULTIHOP], may be needed.

   A device could have three kinds of failures:

        i) Control Channel Failure

       ii) Data Channel Failure

      iii) Control and Data Channel Failure

   The protocol described in this document specifies the recovery in
   conditions i) and iii).  It is perceived that not much (stateful
   information) could be recovered via a control protocol exchange in
   case of ii).

   The failover protocol consists of three phases:

   1) Failover Capability Negotiation: An active endpoint and a remote
      endpoint exchange failover capabilities and attributes to be used
      during the recovery process.

   2) Failover Recovery: A recovery endpoint establishes a new L2TP
      control connection (called recovery tunnel) for every old tunnel
      that it wishes to recover.  The recovery tunnel serves three
      purposes:

      - It identifies the old tunnel that is being recovered.

      - It provides a means of authentication and a three-way handshake
        to ensure both ends agree on the failover for the specified old
        tunnel.




Jain, et al.                Standards Track                     [Page 6]

RFC 4951                        FAILOVER                     August 2007


      - It could exchange the Ns and Nr values to be used in the
        recovered tunnel.

      Upon establishing the recovery tunnel, two endpoints reset the
      control and data channel(s) on the recovered tunnel using the
      procedures described in Section 3.2.2 and Section 3.2.3,
      respectively.  The recovery tunnel could be torn down after that,
      and sessions that were established resume traffic.

   3) Session State Synchronization: The session state synchronization
      process occurs on the recovered or the old tunnel and allows the
      two endpoints to agree on the state of the various sessions in the
      tunnel after failover.  The inconsistency, which could arise due
      to the failure, is handled in the following manner: first, the two
      endpoints silently clear the sessions that were not in the
      established state.  Then, they utilize Failover Session Query
      (FSQ) and Failover Session Response (FSR) on the recovered tunnel
      to obtain the state of sessions as known to the peer endpoint and
      clear the sessions accordingly.

3.  Failover Protocol



   The protocol consists of three steps describing specifications during
   the life of a control connection - before and after failover.

3.1.  Failover Capability Negotiation



   The active and remote endpoints exchange the Failover Capability
   attribute-value pair (AVP) in Start-Control-Connection-Request
   (SCCRQ) and Start-Control-Connection-Reply (SCCRP) during control
   connection establishment as a part of the normal (before failover)
   operation.  The Failover Capability AVP, defined in Section 5.1,
   allows an endpoint to specify if it is control and/or data channel
   failover capable and the time allowed for the recovery for the
   tunnel.

3.2.  Failover Recovery Procedure



   The Failover Recovery Procedure described in this section is
   performed only if there was a control channel failure.  The selection
   of the tunnels to be recovered is implementation specific.

   The Failover Recovery Procedure consists of following three steps,
   which are described in detail in the subsections below:

      - Recovery tunnel establishment

      - Control channel reset



Jain, et al.                Standards Track                     [Page 7]

RFC 4951                        FAILOVER                     August 2007


      - Data channel reset

3.2.1.  Recovery Tunnel Establishment



   The recovery endpoint establishes a new control connection, called
   recovery tunnel, for every old tunnel it wishes to recover.  The
   purpose of the recovery tunnel is solely to recover the corresponding
   old tunnel.  There is a one to one relationship between recovery
   tunnel and recovered/old tunnel

   Recovery tunnel establishment considerations:

      - An LCCE MUST follow the procedures described in [L2TPv2] or
        [L2TPv3] to establish the recovery tunnel.

      - The recovery tunnel MUST use the same L2TP version (and
        establishment procedures) that was used for the old tunnel.

      - The SCCRQ for Recovery tunnel MUST include the Tunnel Recovery
        AVP, defined in Section 5.2, to identify the old tunnel that is
        being recovered.

      - The recovery tunnel MUST NOT include the Failover Capability AVP
        in its SCCRQ or SCCRP messages.

      - An endpoint SHOULD NOT send any message other than the following
        on the recovery tunnel: SCCRQ, SCCRP, SCCCN, StopCCN, HELLO,
        ZLB, and ACK ([L2TPv3] only).

      - An endpoint MUST NOT use any old Tunnel ID for the recovery
        tunnel.  The old tunnels MUST be valid until the recovery
        process concludes.

      - An endpoint MUST use the Tie Breaker AVP (Section 4.4.3
        [L2TPv2]) or Control Connection Tie Breaker AVP (Section 5.4.3
        [L2TPv3]) in the setup of the recovery tunnel to ensure that
        only a single recovery tunnel (when both endpoints have
        simultaneous failover) is established to recover an old tunnel.
        The tunnel that wins the tie is used to decide the suggested Ns
        and Nr values on the recovered tunnel.  Therefore, the endpoint
        that loses the tie, should reset the Ns and Nr values (Section
        3.2.2) as if it were a remote endpoint.  Appendix B illustrates
        the double failover scenario.

      - Tie Breaker AVP processing: The scope of a tie breaker AVP's
        action for recovery and non recovery tunnels must be
        independent, and is defined as follows:




Jain, et al.                Standards Track                     [Page 8]

RFC 4951                        FAILOVER                     August 2007


        o  When Tie Breaker AVP is used in a non recovery tunnel, the
           scope of the tie breaker AVP's action MUST only be within non
           recovery tunnels.  Therefore, losing a tie against a non
           recovery tunnel MUST NOT result in termination of any
           recovery tunnel.

        o  When a Tie Breaker AVP is used in a recovery tunnel, the
           scope of tie breaker AVP's action is further restricted to
           the recovery tunnel(s) for a single tunnel to be recovered.
           Thus, an implementation MUST apply the tie breaker received
           in a recovery tunnel only to those tunnels that are a)
           recovery tunnels, and b) associated with the same tunnel to
           be recovered.  It MUST NOT impact the operation of non-
           recovery tunnels and recovery tunnels associated with other
           old tunnels to be recovered.

   Upon getting an SCCRQ with a Tunnel Recovery AVP, an endpoint
   validates the Recover Tunnel ID and the Recover Remote Tunnel ID and
   responds with an SCCRP.  It MUST terminate the recovery tunnel if:

      - The Recover Tunnel ID or the Recover Remote Tunnel ID is
        unknown.

      - The active or remote endpoint (prior to failover) had not
        indicated that it was failover capable.

      - The L2TP version of recovery tunnel is different from the
        version used in the old tunnel.

   If the remote endpoint accepts the SCCRQ, it SHOULD include the
   Suggested Control Sequence AVP, defined in Section 5.3, in the SCCRP
   message.

   Authentication considerations:

      - To authenticate a peer endpoint during recovery tunnel
        establishment, an endpoint MUST follow the procedure described
        in either [L2TPv2] Section 5.1.1 or [L2TPv3] Section 4.3.  It
        MUST use the same secret that was used to authenticate the old
        tunnel.

      - Not being able to authenticate could be a reason to terminate
        the recovery tunnel.

      - For L2TPv3 tunnels, a recovery tunnel MUST use the Control
        Message authentication (i.e., exchange the nonce values), as
        described in [L2TPv3] Section 4.3, if the old tunnel was
        configured to do control message authentication.  An L2TPv3



Jain, et al.                Standards Track                     [Page 9]

RFC 4951                        FAILOVER                     August 2007


        recovered tunnel MUST reset its nonce values (both endpoints) to
        the nonce values exchanged in the recovery tunnel.

   For any reason, if the recovery endpoint could not establish the
   recovery tunnel, then it MUST silently clear the old tunnel and
   sessions within, concluding that the recovery process has failed.

   Any control packet received on the recovered tunnel before control
   channel reset (Section 3.2.2) MUST be silently discarded.

3.2.2.  Control Channel Reset



   Control channel reset allows new control messages to be sent and
   received over the recovered tunnel.

   Control channel reset procedure:

      - An endpoint SHOULD flush the transmit/receive windows and reset
        the control channel sequence numbers (i.e., Ns and Nr values) on
        the recovered tunnel.  The control channel on the recovery
        endpoint is reset upon getting a valid SCCRP on the recovery
        tunnel, whereas the control channel on the remote endpoint is
        reset upon getting a valid SCCCN on the recovery tunnel.  If the
        recovery endpoint did not receive the Suggested Control Sequence
        (SCS) AVP in the SCCRP then it MUST reset the Ns and Nr values
        to zero.  If the remote endpoint opted to not send the SCS AVP
        then it MUST reset the Ns and Nr values to zero.  Either
        endpoint can tear down the recovery tunnel after the control
        channel reset procedure is complete.

      - An endpoint MUST prevent the establishment of new sessions until
        it has cleared (or marked for clearance) the sessions that were
        not in an established state, i.e., until after Step I, Section
        3.3 is complete.

3.2.3.  Data Channel Reset



   A data channel reset procedure is applicable only for the sessions
   using sequence numbers.  For L2TPv3 data channel, the terms Nr and Ns
   in this document are used to mean "expected sequence number" and
   "sequence number", respectively.

   Data channel reset procedure:

      - The recovery endpoint sets the Ns value to zero.

      - The remote endpoint (recovery endpoint's peer) continues to use
        the Ns values it was using previously.



Jain, et al.                Standards Track                    [Page 10]

RFC 4951                        FAILOVER                     August 2007


      - To reset Nr values during failover, if an endpoint receives 'n'
        out of order but in sequence packets, then it MUST set the Nr
        value based on the Ns value of the incoming packets, as
        suggested in Appendix C of [L2TPv3].  The value of 'n' SHOULD be
        configurable.

      - If one of the endpoints does not exhibit the capability
        (indicated in 'D' bit in the Failover Capability AVP) to reset
        the Nr value, then data channels using sequence numbers are
        considered non recoverable.  Those sessions SHOULD be torn down
        by the recovery endpoint by sending a Call-Disconnect-Notify
        (CDN).

      - For data-channel-only failure, two endpoints MAY use the session
        state query/response mechanism on the control channel to
        synchronize the state of sessions as described in Section 3.3
        below.

3.3.  Session State Synchronization



   If a control channel failure happens when a session was being
   established or torn down, then it is possible for an endpoint to
   consider a session in an established state while its peer considers
   the same session non existent.  Two such situations occur when
   failure on an endpoint occurs immediately after sending:

      - A CDN message that never made it to the peer.

      - An ICCN message that never made it to the peer.

   The following mechanism MUST be used to identify and clear the
   sessions that exists on an endpoint, but not on its peer:

   Step I: For control channel failure, after the recovery tunnel is
   established, the sessions that were not in an established state MUST
   be silently cleared (i.e., without sending a CDN message) by each
   endpoint.

   Step II: Both endpoints MAY identify the sessions that might have
   been in inconsistent states, perhaps based on data channel
   inactivity.  FSQ and FSR messages have been introduced to synchronize
   session state at any given point during the life of a session between
   two endpoints.  These messages are used when one endpoint determines
   or suspects in an implementation specific manner that its session
   state could be inconsistent with that of its peer's.

   Step III: An endpoint sends a Failover Session Query (FSQ) message to
   query the state of sessions as known to its peer.  An FSQ message



Jain, et al.                Standards Track                    [Page 11]

RFC 4951                        FAILOVER                     August 2007


   contains one Failover Session State (FSS) AVP, defined in Section
   5.4, for each session it wishes to query.  Multiple FSS AVPs could be
   included in one FSQ message.  An FSQ message MUST include at least
   one FSS AVP.  An endpoint MAY send another FSQ message before getting
   a response for its previous FSQs.

   An inconsistency about a session's existence during failover could
   result in an endpoint selecting the same Session ID for a new
   session.  In such a situation, it would send an ICRQ for an already
   established session.  Therefore, before all sessions are synchronized
   using an FSQ/FSR mechanism, if endpoint receives an ICRQ for a
   session in an established state, then it MUST respond to such an ICRQ
   with a CDN.  The CDN message must set Assigned/Local Session ID AVP
   ([L2TPv2] Section 4.4.4, [L2TPv3] Section 5.4.4) to its local Session
   ID and clear the session that it considered established.  Use of a
   least recently used Session ID for the new sessions could help reduce
   this symptom during failover.

   When an endpoint receives an FSQ message, it MUST ensure that for
   each FSS AVP in the FSQ message, it includes an FSS AVP in the
   Failover Session Response (FSR) message.  An endpoint could respond
   to multiple FSQs using one FSR message, or it could respond one FSQ
   with multiple FSRs.  FSSs are not required to be responded in the
   same order in which they were received.  For each FSS AVP received in
   FSQ messages, an endpoint MUST validate the Remote Session ID and
   determine if it is paired with the Session ID specified in the
   message.  If an FSS AVP is not valid (i.e., session is non-existing
   or it is paired with different remote Session ID), then the Session
   ID field in the FSS AVP in the FSR MUST be set to zero.  When session
   is discovered to be pairing with mismatching Session ID, the local
   session MUST not be cleared, but rather marked stale, to be queried
   later using an FSQ message.  Appendix C presents an example dialogue
   between two endpoints with mismatching Session IDs.

   When responding to an FSQ with an FSR message, the Remote Session ID
   in the FSS AVP of the FSR message is always set to the received value
   of the Session ID in the FSS AVP of the FSQ message.

   When an endpoint receives an FSR message, for each FSS AVP it MUST
   use the Remote Session ID field to identify the local session and
   silently (without sending a CDN) clear the session if the Session ID
   in the AVP was zero.  Otherwise, it MUST consider the session to be
   in an established state and recovered.

4.  New Control Messages



   This document introduces two new messages that could be sent over an
   established/recovered control connection.



Jain, et al.                Standards Track                    [Page 12]

RFC 4951                        FAILOVER                     August 2007


4.1.  Failover Session Query



   The Failover Session Query (FSQ) control message is used by an
   endpoint during the recovery process to query the state of various
   sessions.  It triggers a response from the peer, which contains the
   requested state of various sessions.

   This control message is encoded as follows:

      Vendor ID = 0 (IETF)
      Attribute Type = 21

   The following AVPs MUST be present in the FSQ control message:

      Message Type
      Failover Session State

   The following AVPs MAY be present in the FSQ control message:

      Random Vector
      Message digest ([L2TPv3] tunnels only)

   Other AVPs MUST NOT be sent in this control message and SHOULD be
   ignored on receipt.

   The M-bit on the Message Type AVP for this control message MUST be
   set to 0.

4.2.  Failover Session Response



   The Failover Session Response (FSR) control message is used by an
   endpoint during the recovery process to respond with the local state
   of various sessions.  It is sent as a response to an FSQ message.  An
   endpoint MAY choose to respond to an FSQ message with multiple FSR
   messages.

   This control message is encoded as follows:

      Vendor ID = 0 (IETF)
      Attribute Type = 22

   The following AVPs MUST be present in the FSR control message:

      Message Type
      Failover Session State






Jain, et al.                Standards Track                    [Page 13]

RFC 4951                        FAILOVER                     August 2007


   The following AVPs MAY be present in the FSR control message:

      Random Vector
      Message digest ([L2TPv3] tunnels only)

   Other AVPs MUST NOT be sent in this control message and SHOULD be
   ignored on receipt.

   The M-bit on the Message Type AVP for this control message MUST be
   set to 0.

5.  New Attribute Value Pairs



   The following sections contain a list of new L2TP AVPs defined in
   this document.

5.1.  Failover Capability AVP



   The Failover Capability AVP, Attribute Type 76, indicates the
   capabilities of an endpoint required for the recovery process.  The
   AVP format is defined as follows:

   Failover Capability AVP

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |                0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Attribute Type 76     |         Reserved          |D|C|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |              Recovery Time (in milliseconds)                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The AVP MAY be hidden (the H-bit set to 0 or 1).  The AVP is not
   mandatory (the M-bit MUST be set to 0).

   The C bit governs the failover capability for the control channel.
   When the C bit is set, it indicates that the endpoint can recover
   from a control channel failure using the procedure described in
   Section 3.2.2.

   When the C bit is not set, it indicates that the endpoint cannot
   recover from a control channel failover.  In this case, the D bit
   MUST be set.  Note that a control channel failover in this case would
   be fatal for the tunnel and all associated data channels.





Jain, et al.                Standards Track                    [Page 14]

RFC 4951                        FAILOVER                     August 2007


   The D bit governs the failover capability for data channels that use
   sequence numbers.  Data channels that do not use sequence numbers do
   not need help to recover from a data channel failure.

   When the D bit is set, it indicates that the endpoint is capable of
   resetting Nr value of data channels using the procedure described in
   Section 3.2.3 Data Channel reset procedure.

   When the D bit is not set, it indicates that the endpoint cannot
   recover data channels that use sequence numbers.  In the case of a
   failure, such data channels would be lost.

   The Failover Capability AVP MUST NOT be sent with C bit and D bit
   cleared.

   The Recovery Time, applicable only when the C bit is set, is the time
   in milliseconds an endpoint asks its peer to wait before assuming the
   recovery process has failed.  This timer starts when an endpoint's
   control channel timeout ([L2TPv2] Section 5.8, [L2TPv3] Section 4.2)
   is started, and is not stopped (before expiry) until an endpoint
   successfully authenticates its peer during recovery.  A value of zero
   does not mean that failover will not occur, it means no additional
   time is requested from the peer.  The timer is also stopped if a
   control channel message is acknowledged by the peer in the situation
   when there was no failover, but the loss of the control channel
   message was a temporary phenomenon.

   This AVP MUST NOT be included in any control message other than SCCRQ
   and SCCRP messages.

5.2.  Tunnel Recovery AVP



   The Tunnel Recovery AVP, Attribute Type 77, indicates that a sender
   would like to recover the tunnel identified in this AVP due to a
   failure.  The AVP format is defined as follows:
















Jain, et al.                Standards Track                    [Page 15]

RFC 4951                        FAILOVER                     August 2007


   Tunnel Recovery AVP for L2TPv3 tunnels:

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |                0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Attribute Type 77     |           Reserved            |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Recover Tunnel ID                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Recover Remote Tunnel ID                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


   Tunnel Recovery AVP for L2TPv2 tunnels:

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |                0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Attribute Type 77     |           Reserved            |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Reserved              |     Recover Tunnel ID         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Reserved              |   Recover Remote Tunnel ID    |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   This AVP MUST not be hidden (the H-bit is set to 0).  The AVP is
   mandatory (the M-bit is set to 1).

   The Recover Tunnel ID encodes the local Tunnel ID that an endpoint
   wants recovered.  The Recover Remote Tunnel ID encodes the remote
   Tunnel ID corresponding to the old tunnel.

   This AVP MUST NOT be included in any control message other than the
   SCCRQ message when establishing a Recovery Tunnel.

5.3.  Suggested Control Sequence AVP



   The Suggested Control Sequence (SCS) AVP, Attribute Type 78,
   specifies the Ns and Nr values to for the recovered tunnel.  This AVP
   is included in an SCCRP message of a recovery tunnel by remote
   endpoint.  The AVP format is defined as follows:






Jain, et al.                Standards Track                    [Page 16]

RFC 4951                        FAILOVER                     August 2007


    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |                0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Attribute Type 78     |            Reserved           |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |        Suggested Ns           |         Suggested Nr          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   This AVP MAY be hidden (the H-bit set to 0 or 1).  The AVP is not
   mandatory (the M-bit is set to 0).

   This is an optional AVP, suggesting Ns and Nr values to be used by
   the recovery endpoint.  If this AVP is present in an SCCRP message
   during recovery tunnel establishment, the recovery endpoint MUST set
   the Ns and Nr values of the recovered tunnel to the respective
   suggested values.  When this AVP is not sent in an SCCRP or not
   present in an incoming SCCRP, the Ns and Nr values for the recovered
   tunnel are set to zero.  Use of this AVP helps avoid the interference
   in the recovered tunnel's control channel with old control packets.

   This AVP MUST NOT be included in any control message other than the
   SCCRP message when establishing a Recovery Tunnel.

5.4.  Failover Session State AVP



   The Failover Session State (FSS) AVP, Attribute Type 79, is used to
   query the state of a session from the peer end to clear the sessions
   that otherwise would remain in an undefined state after failover.
   The AVP format is defined as follows:

   FSS AVP format for L2TPv3 sessions:

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |                0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |      Attribute Type 79        |         Reserved              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Session ID                           |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                       Remote Session ID                       |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+






Jain, et al.                Standards Track                    [Page 17]

RFC 4951                        FAILOVER                     August 2007


   FSS AVP format for L2TPv2 sessions:

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |                0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |      Attribute Type 79        |         Reserved              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |            Reserved           |        Session ID             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |            Reserved           |      Remote Session ID        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   This AVP MAY be hidden (the H-bit set to 0 or 1).  The AVP is
   mandatory (the M-bit is set to 1).

   The Session ID identifies the local Session ID that the sender had
   assigned, for which it would like to query the state on its peer.  A
   Remote Session Id is the remote Session ID for the same session.

   An FSS AVP MUST NOT be used in any message other than FSQ and FSR
   messages.

6.  Configuration Parameters



   An L2TP endpoint MAY expose the following configuration parameters to
   be specified for control connections:

      - Control Channel Failover Capability: Failover Capability AVP
        (Section 5.1), C bit.

      - Data Channel Failover Capability: Failover Capability AVP
        (Section 5.1), D bit.

      - Recovery Time: Failover Capability AVP (Section 5.1).

   The L2TP MIB defined in [L2TPv2-MIB] and [L2TPv3-MIB], defines a
   number of objects that may be used for monitoring the status L2TP
   nodes, but is seldom used for configuration purposes.  It is expected
   that the above mentioned parameters will be configured by using a
   Command Line Interface (CLI) or other proprietary mechanism.

   Asynchronous notifications for failover and recovery events may be
   sent by L2TP nodes to network management applications, but the
   specification of the protocol and format to be used for these
   notifications is out of the scope of this document.




Jain, et al.                Standards Track                    [Page 18]

RFC 4951                        FAILOVER                     August 2007


7.  IANA Considerations



   This document defines the following values assigned by IANA.

   - Four Control Message Attribute Value Pairs (Section 10.1 [L2TPv3]):

           Failover Capability         : 76
           Tunnel Recovery             : 77
           Suggested Control Sequence  : 78
           Failover Session State      : 79


   - Two Message Type (Attribute Type 0) Values (Section 10.2 [L2TPv3]):

           Failover Session Query      : 21
           Failover Session Response   : 22

8.  Security Considerations



   A spoofed failover request (SCCRQ with Tunnel Recovery AVP) on behalf
   of an endpoint might cause a control channel termination if
   authentication measures mentioned in Section 3.2.1 are not used.

   Even if the authentication measures (as described in Section 3.2.1)
   were used, it is still possible to learn an identity of an
   operational tunnel from an endpoint by issuing it spoofed failover
   requests that fail the authentication procedure.  The probability of
   succeeding with a spoofed failover request is 1 in (2^16 - 1) for
   [L2TPv2] and 1 in (2^32 - 1) for [L2TPv3].  The discovered identity
   of an operational tunnel could then be misused to send control
   messages for a possible hindrance to the control connection.
   Typically, control messages that are outside the endpoint's receive
   window are discarded.  However, if Suggested Control Sequence AVP
   (Section 5.3) is not used during the actual failover process, the
   sequence numbers might be reset to zero, thereby making the receive
   window predictable.  To improve security under such circumstances, an
   endpoint may be configured with the possible set of recovery
   endpoints that could recover a tunnel, and use of Suggested Control
   Sequence AVP when recovering a tunnel.

9.  Acknowledgements



   Leo Huber provided suggestions to help define the failover concept.
   Mark Townsley, Carlos Pignataro, and Ignacio Goyret reviewed the
   document and provided valuable suggestions.






Jain, et al.                Standards Track                    [Page 19]

RFC 4951                        FAILOVER                     August 2007


10.  Contributors



   Paul Howard            Juniper Networks
   Vipin Jain             Riverstone Networks
   Sam Henderson          Cisco Systems
   Keyur Parikh           Harris Corporations

11.  References



11.1.  Normative References



   [RFC2119]      Bradner, S., "Key words for use in RFCs to Indicate
                  Requirement Levels", BCP 14, RFC 2119, March 1997.

   [L2TPv2]       Townsley, W., Valencia, A., Rubens, A., Pall, G.,
                  Zorn, G., and B. Palter, "Layer Two Tunneling Protocol
                  "L2TP"", RFC 2661, August 1999.

   [L2TPv3]       Lau, J., Townsley, M., and I. Goyret, "Layer Two
                  Tunneling Protocol - Version 3 (L2TPv3)", RFC 3931,
                  March 2005.

11.2.  Informative References



   [L2TPv2-MIB]   Caves, E., Calhoun, P., and R. Wheeler, "Layer Two
                  Tunneling Protocol "L2TP" Management Information
                  Base", RFC 3371, August 2002.


   [L2TPv3-MIB]   Nadeau, T. and K. Koushik, "Layer Two Tunneling
                  Protocol (version 3) Management Information Base",
                  Work in Progress, August 2006.

   [BFD-MULTIHOP] Katz, D. and D. Ward, "BFD for Multihop Paths", Work
                  in Progress, March 2007.
















Jain, et al.                Standards Track                    [Page 20]

RFC 4951                        FAILOVER                     August 2007


Appendix A

   Description below outlines the failover protocol operation for an
   example tunnel.  The failover protocol does not preclude an endpoint
   from recovering multiple tunnels in parallel.  It also allows an
   endpoint to send multiple FSQs, each including multiple FSS AVPs, to
   recover quickly.

   Failover Capability Negotiation (Section 3.1):

  Endpoint                                             Peer
               (assigned tid = x, failover capable)
  SCCRQ       -------------------------------------->  validate SCCRQ

               (assigned tid = y, failover capable)
  validate    <--------------------------------------  send SCCRP
  SCCRP, etc.

   .... <after tunnel gets created, sessions are established> ....


  < This Node fails >

  The Recovery endpoint establishes the recovery tunnel (Section 3.2.1).
  Initiate recovery tunnel establishment for the old tunnel 'x':

  Recovery Endpoint                                     Peer

            (assigned tid = z, Recovery AVP)
  SCCRQ     ----------------------------------->  Detects failover
          (recover tid = x, recover remote tid = y)  validate SCCRQ


          (Suggested Control Sequence AVP, Suggested Ns/Nr = 3/100)
  validate <-----------------------------------   send SCCRP
  SCCRP    (recover tid = y, recover remote tid = x)
  reset Ns = 3, Nr = 100
  on the recovered tunnel

  SCCCN     ----------------------------------->  validate and reset
                                                  Ns = 100, Nr = 3 on
                                                  the recovered tunnel









Jain, et al.                Standards Track                    [Page 21]

RFC 4951                        FAILOVER                     August 2007


  Terminate the recovery tunnel

  tid = 'z'
  StopCCN  --------------------------------------> Cleanup 'w'


  Session states are synchronized both endpoints may send FSQs and
  cleanup stale sessions (Section 3.3)

             (FSS AVP for sessions s1, s2, s3..)
  send FSQ  -------------------------------------> compute the state
                                                      of sessions in FSQ

                (FSS AVP for sessions s1, s2, s3...)
     deletes  <-------------------------------------- send FSR
     stale sessions, if any


                (FSS AVP for sessions s7, s8, s9...)
     compute  <-------------------------------------- send FSQ
     the sate of
     sessions in FSQ


                (FSS AVP for sessions s7, s8, s9...)
     send FSR --------------------------------------> delete stale
                                                      sessions, if any
























Jain, et al.                Standards Track                    [Page 22]

RFC 4951                        FAILOVER                     August 2007


Appendix B

   This section shows an example dialogue to illustrate double failure
   recovery.  The notable difference, as described in Section 3.2.1, in
   the procedure from single failover scenario is the use of a tie
   breaker by one of the recovery endpoints to use the recovery tunnel
   established by its peer (also a recovery endpoint) as a recovery
   tunnel.

      Recovery endpoint                     Recovery endpoint

      (assume old tid = A)                 (assume old tid = B)

                  Recovery AVP = (A, B)
      SCCRQ     -----------------------+
      (with tie  (recovery tunnel 'C') |
       breaker                         |
       AVP)                            |
                 Recovery AVP = (B, A) |
   +- valid    <--------------------------- Send SCCRQ
   |  SCCRQ      (recovery tunnel 'D') |    (with tie breaker AVP)
   |  This endpoint                    |
   |  loses tie;                       |
   |  Discards tunnel 'C'              +--> Valid SCCRQ
   |                                        This endpoint wins tie;
   |                                        Discards SCCRQ
   |
   |              (may include SCS AVP)
   +->Send SCCRP -------------------------> Validate SCCRP
                                            Reset 'B';
                                            Set Ns, Nr values --+
                                                                |
                                                                |
                                                                |
      Validate SCCN <---------------------- Send SCCN    -------+
      Reset 'A';
      Set Ns, Nr values

   FSQs and FSRs for the old tunnel (A, B) are exchanged on the
   recovered tunnel by both endpoints.











Jain, et al.                Standards Track                    [Page 23]

RFC 4951                        FAILOVER                     August 2007


Appendix C

   Session ID mismatch could not be a result of failure on one of the
   endpoints.  However, failover session recovery procedure could
   exacerbate the situation, resulting into a permanent mismatch in
   Session IDs between two endpoints.  The dialogue below outlines the
   behavior described in Section 3.3, Step III to handle such situations
   gracefully.

   Recovery endpoint                    Remote endpoint

   (assume a mismatch)                  (assume a mismatch)
   Sid = A, Remote Sid = B              Sid = B, Remote Sid = C
   Sid = C, Remote Sid = D


                  FSS AVP (A, B)
   send FSQ  -------------------------> No (B, A) pair exist;
                                        rather (B, C) exist.
                                        If it clears B then peer doesn't
                                        know if C is stale on other end.

                                        Instead if it marks B stale
                                        and queries the session state
                                        via FSQ, C would be cleared on
                                        the other end.

                  FSS AVP (0, A)
   Clears A <-------------------------- send FSR

                                        ... some time later ...

                  FSS AVP (B, C)
   No (C,B) <-------------------------- send FSQ
   Mark C Stale

                  FSS AVP (0, B)
   Send FSR --------------------------> Clears B













Jain, et al.                Standards Track                    [Page 24]

RFC 4951                        FAILOVER                     August 2007


Author Information

   Vipin Jain
   Riverstone Networks
   5200 Great America Parkway
   Santa Clara, CA 95054
   EMail: vipinietf@yahoo.com

   Paul W. Howard
   Juniper Networks
   10 Technology Park Drive
   Westford, MA 01886
   EMail: phoward@juniper.net

   Sam Henderson
   Cisco Systems
   7025 Kit Creek Rd.
   PO Box 14987
   Research Triangle Park, NC 27709
   EMail: samh@cisco.com

   Keyur Parikh
   Harris Corporation
   4393 Digitalway
   Mason, OH 45040
   EMail: kparikh@harris.com

























Jain, et al.                Standards Track                    [Page 25]

RFC 4951                        FAILOVER                     August 2007


Full Copyright Statement



   Copyright (C) The IETF Trust (2007).

   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.

   This document and the information contained herein are provided on an
   "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
   OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND
   THE INTERNET ENGINEERING TASK FORCE DISCLAIM 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.

Intellectual Property



   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
   http://www.ietf.org/ipr.

   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
   ietf-ipr@ietf.org.

Acknowledgement



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







Jain, et al.                Standards Track                    [Page 26]