Internet DRAFT - draft-cheng-nsis-multihoming

draft-cheng-nsis-multihoming






Next Steps in Signaling (nsis)                                  H. Cheng
Internet-Draft                                                  T. Sanda
Intended status: Standards Track                                   T. Ue
Expires: August 5, 2007                                       S. Marwaha
                                                               Panasonic
                                                                Feb 2007


                    NSIS Multihoming Support Issues
                  draft-cheng-nsis-multihoming-00.txt

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-
   Drafts.

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

   The list of current Internet-Drafts can be accessed at
   http://www.ietf.org/ietf/1id-abstracts.txt.

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

   This Internet-Draft will expire on August 5, 2007.

Copyright Notice

   Copyright (C) The IETF Trust (2007).












Cheng, et al.            Expires August 5, 2007                 [Page 1]

Internet-Draft              NSIS Multihoming                    Feb 2007


Abstract

   The NSIS operation will be greatly affected by the multihoming
   development in IETF such as Monami6, Shim6, etc.  Solutions from
   these groups introduce different architecture elements and
   assumptions to the network layer, and thus render the current NSIS
   protocols ineffective or irrelevant.  This draft provides an analysis
   of the multhoming scenarios that have impacts on the NSIS operation.
   Requirements for NSIS multihoming support are derived from the
   discussions of the NSIS applicability in these scenarios.


Table of Contents

   1.  Conventions used in this document  . . . . . . . . . . . . . .  3
   2.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  4
   3.  Terminology  . . . . . . . . . . . . . . . . . . . . . . . . .  5
   4.  NSIS Applicability on Static Multihoming scenarios . . . . . .  6
     4.1.  Single CoA with multiple HoAs  . . . . . . . . . . . . . .  6
       4.1.1.  Using bi-direction tunnel via home agents  . . . . . .  6
       4.1.2.  Using Route Optimization . . . . . . . . . . . . . . .  8
     4.2.  Multiple CoAs with single HoA  . . . . . . . . . . . . . . 10
       4.2.1.  Using bi-directional tunnel via home agents  . . . . . 10
       4.2.2.  Using Route Optimization . . . . . . . . . . . . . . . 12
     4.3.  Multiple CoAs with Multiple HoAs . . . . . . . . . . . . . 14
   5.  Multihoming Scenarios with mobility  . . . . . . . . . . . . . 15
     5.1.  Single CoA with multiple HoAs  . . . . . . . . . . . . . . 15
       5.1.1.  Using bi-direction tunnel via home agents  . . . . . . 15
       5.1.2.  Using Route Optimization . . . . . . . . . . . . . . . 15
     5.2.  Multiple CoAs with single HoA  . . . . . . . . . . . . . . 15
       5.2.1.  Using bi-direction tunnel via home agents  . . . . . . 16
       5.2.2.  Using Route optimization . . . . . . . . . . . . . . . 18
     5.3.  Multiple CoAs with a single HoA  . . . . . . . . . . . . . 18
   6.  Problem statement for NSIS multihoming support . . . . . . . . 19
   7.  Security Considerations  . . . . . . . . . . . . . . . . . . . 20
   8.  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 21
   9.  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 22
   10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 23
     10.1. Normative Reference  . . . . . . . . . . . . . . . . . . . 23
     10.2. Informative References . . . . . . . . . . . . . . . . . . 23
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 24
   Intellectual Property and Copyright Statements . . . . . . . . . . 25









Cheng, et al.            Expires August 5, 2007                 [Page 2]

Internet-Draft              NSIS Multihoming                    Feb 2007


1.  Conventions used in this document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC2119 [1].














































Cheng, et al.            Expires August 5, 2007                 [Page 3]

Internet-Draft              NSIS Multihoming                    Feb 2007


2.  Introduction

   Support of multihoming has become essential to the success of a
   protocol in the mobile environment.  With the integration of access
   technologies and proliferation of multimode terminals, multihoming
   scenarios are commonplace in modern networks.  Naturally, use of
   multihoming is viewed as a solution to provide ubiquitous and fault
   tolerant access to the Internet with better user experience.  Several
   groups in IETF have set out to develop solutions on multihoming in
   response to the market demand, e.g.Monami6 [2], Shim6 [3], etc.
   However NSIS WG has yet to come up with an analysis of the impact of
   these new developments.

   New solutions developed in these WGs address only the data bearer
   aspect of the multihoming.  The signaling control aspect, the area
   related to NSIS, is left unexplored.  Although data bearer solutions
   conceal multihoming details from upper layers, path coupled signaling
   control still needs to work out the paths individually, e.g. multiple
   end-to-end QoS NSLP signaling have to be carried out on all paths for
   a load sharing case.  Therefore, NSIS faces a more vibrant and
   complicated environment when multihoming solutions are used, and
   difference requirements would be placed on its operations.

   This draft studies the multihoming scenarios and identifies operation
   issues relevant to NSIS, i.e. multiple paths are used simultaneous
   for a session.  Requirements for the NSIS multihoming support are
   derived from the discussion.  As there are different multihoming
   solutions in development, this draft does not concentrate on a
   specific scheme.  Instead, the scenarios studied are loosely based on
   a more generic classification [9].  Requirements resulted from the
   study could be used by the WG in defining necessary changes and
   enhancement for NSIS protocols to support multihoming.



















Cheng, et al.            Expires August 5, 2007                 [Page 4]

Internet-Draft              NSIS Multihoming                    Feb 2007


3.  Terminology

   The Terminology defined in [4], applies to this draft.  In addition,
   the following terms are used:

   CCRN: CN side CRN.  Assuming MN is the multihoming node and CN is the
   normal node, CCRN is the point where the multiple paths split.  In
   other words, between CN and CCRN, all the flows share the common
   path.

   MCRN: MN side CRN.  This is the point where multiple paths join
   before reaching MN.  It should only exist when MN has a single CoA or
   multiple CoA belonging to the same subnet.

   HCRN: Home Agent side CRN.  Assuming MN is the multihoming node, HCRN
   is the point where the multiple paths split when reverse tunneling is
   used.  Between HA and the MCRN (in turn between CN and MCRN), all the
   flows share the common path.

   TCRN: Transient CRN.  This is a crossover node between the new path
   and the old path due to the MN's mobility.  It is equivalent to the
   normal CRN definition in Mobility Applicability I-D [5].  In certain
   scenarios, the TCRN could collocate with the CCRN, MCRN, or HCRN.  In
   other scenarios, the TCRN could be different from either of the other
   CRNs.


























Cheng, et al.            Expires August 5, 2007                 [Page 5]

Internet-Draft              NSIS Multihoming                    Feb 2007


4.  NSIS Applicability on Static Multihoming scenarios

   This section analyzes NSIS applicability on various multihoming
   scenarios based on current NSIS specifications.

   IP level multihoming could result from varies reasons.  For example,
   a node may use the site multihoming solution to achieve redundancy or
   load sharing [6].  A multimode terminal may also access services
   simultaneously via both interfaces.

   However, not all IP level multihoming scenarios require special
   functions for the path-coupled signaling, e.g.  NSIS.  For example,
   if an application session only makes use of one fixed path, it
   obviously has no difference from a normal single-homed scenario.

   Therefore, in the following scenario analysis, focus is placed on
   cases where the different paths are used for the same session or
   related sessions.

   The scenarios are presented using Mobile IPv6 related cases as
   examples.  However, this does not imply the issues are only limited
   to MIPv6.  Similar issue exists even for fixed nodes communications.
   The relevance to fixed node cases is covered in detail analysis in
   each subsection.

4.1.  Single CoA with multiple HoAs

   This scenario happens when a single interface device has multiple
   subscription profiles and tries to use them all for the same
   application session.  Another possible case is when a multiple
   interface device has one link down, and makes use of the live
   interface as if it roamed.

4.1.1.  Using bi-direction tunnel via home agents

   In this scenario, when current NSIS operation procedure is followed,
   e.g.  QoS NSLP [4] and NSIS Tunneling [7], session binding
   information may get lost over the path from different HAs to the MN.
   Details of the issue are presented with the example below.

   Figure 1 shows an example of the scenario and the possible data
   paths.  A MN having two HoAs, HoA-a and HoA-b obtained from HA1 and
   HA2 respectively, is communicating with a CN via the bi-directional
   tunnel through the Home Agents, i.e.  HA1 and HA2.  Data path from CN
   towards the two Home Agents splits at CCRN; and tunneled data paths
   from the two Home Agents to the MN merges at MCRN.





Cheng, et al.            Expires August 5, 2007                 [Page 6]

Internet-Draft              NSIS Multihoming                    Feb 2007


                  +--+
                  |CN|
                  +--+
                   ab
                   ab
                 +----+
          aaaaaaa|CCRN|bbbbbbb
          a      +----+      b
          a                  b
        +---+              +---+
        |HA1|              |HA2|
        +---+              +---+
          A                  B
          A      +----+      B
          AAAAAAA|MCRN|BBBBBBB
                 +----+
                   AB
                   AB
                  +--+
                  |AR|        aaaaa  Path between CN and HA1
                  +--+        AAAAA  Bi-directional tunnel between
                   AB                   MN and HA1
                   AB
                  +--+        bbbbb  Path between CN and HA2
                  |MN|        BBBBB  Bi-directional tunnel between
                  +--+                  MN and HA2



   Figure 1.  CoA, Multiple HoA, with bi-directional tunnel

   Obviously, as shown in the figure, the two home addresses of the MN
   are visible to the CN.  Therefore, multihoming schemes are necessary
   at CN to support the communication, e.g.  SCTP or Shim6.

   When applying NSIS signaling over the data paths, two different End-
   to-End (E2E) flows should be used, due to the two home addresses (of
   MN) and thus two paths towards MN.  Between CN and HA1, FlowID-a is
   used; and between CN and HA2, FlowID-b is used.  As for the Session
   Identifier, one has the choice of using a single SID or different
   SIDs.

   In the above scenario, when the signaling initiator is CN, e.g.  CN
   is QNI, binding of the two flows is problematic between the MCRN and
   MN if operation specified in Tunneling Draft I-D [7] is followed.

   When two different SIDs were used, e.g.  SID-a and SID-b, a
   BOUND_SESSION_ID would be included into the E2E signaling messages.



Cheng, et al.            Expires August 5, 2007                 [Page 7]

Internet-Draft              NSIS Multihoming                    Feb 2007


   For example, message sent to HA1 is using a Session Identifier SID-a,
   and having a BOUND_SESSION_ID of value SID-b.  Therefore, over the
   common path, i.e. from CN to CCRN, the two sessions would be bound as
   desired.  However, when the E2E message reaches the home agents,
   tunneling operation would apply.  According to Tunneling Draft I-D
   [7], a new tunnel session, say SID-A, will be created at HA1; and the
   E2E signaling message would be tunneled towards the tunnel end point,
   i.e.  MN, with a BOUND_SESSION_ID object set to SID-A included.
   Similarly, at HA2, a new tunnel session SID-B will be created for the
   E2E signaling session SID-b, and the E2E signaling message would be
   tunneled towards MN with a BOUND_SESSION_ID object set to SID-B.
   Obviously, the two sessions created for the tunnel sections are no
   longer bound, since no BOUND_SESSION_ID is carried in the tunnel
   session messages.  Moreover, there is no existing method for the two
   tunnel entry points, i.e.  HA1 and HA2, to exchange the new SID
   information.  Therefore, even if BOUND_SESSION_ID object is allowed
   for tunnel session messagess, no correct value could be set.

   When a single SID was used for the two E2E flows, e.g.  SID-ab, no
   BOUND_SESSION_ID object would be necessary for the E2E flow
   signaling.  However, proper flags need to be set to indicate their
   relationships.  For example, for QoS NSLP [4] messages, the REPLACE
   flag should be unset in messages of both flows.  Similarly, at the
   home agents, if new sessions are created for the tunnel paths, no
   BOUND_SESSION_ID would be present in the tunnel signaling.  This
   again leaves the two tunnel sessions unbound.

   When generalized, the above problem could be state as: two flows
   would lose their binding relationship if both go through tunnels via
   different tunnel entry points.  This would be a major issue for
   multihoming devices, where using multiple paths for load sharing is
   common.

4.1.2.  Using Route Optimization

   In this scenario, it is possible that different flows share the same
   MRI after address swapping for Route Optimization, i.e. original
   flows are only differentiated by home address.  In this case, NSIS
   needs to interact with Mobile IP stack to trigger aggregation when
   necessary, so that signaled specification could be enforceable.
   Following example provides further discussion of the issue.

   Figure 2 shows the scenario when the CN support Route Optimization
   (RO).  Since the MN has only one CoA, IP address alone would not
   separate the two flows.  Therefore, CCRN and MCRN only exist if
   routing is based on higher layer information, e.g. port number, etc.





Cheng, et al.            Expires August 5, 2007                 [Page 8]

Internet-Draft              NSIS Multihoming                    Feb 2007


                +--+
                |CN|
                +--+
                 cd
                 cd
                 cd
               +----+
        ccccccc|CCRN|ddddddd
        c      +----+      d
        c                  d
        c                  d
        c                  d
        c      +----+      d
        ccccccc|MCRN|ddddddd
               +----+
                 cd
                 cd
                +--+
                |AR|        ccccc  RO Path for HoA-a
                +--+
                 cd         ddddd  RO Path for HoA-b
                 cd
                +--+
                |MN|
                +--+



   Figure 2.  Single CoA, Multiple HoAs, with Route Optimization

   In certain cases, the two flows may have identical MRIs.  For
   example, CN and MN use the same higher layer port number for the
   application.  This is highly possible for applications based on
   connectionless transport, e.g.  UDP.  For such cases, QNEs on the
   path will have difficulty to distinguish packets from the two flows,
   since the normally used classifier does not check the Type 2 Routing
   Headers.

   Obviously, aggregation is necessary at the end host since the
   intermediary QNEs has no way to enforce QoS for the flows separately.
   Therefore, NSLP layer signaling application needs to monitor the MRI
   in use at the end hosts.  When the same MRI is used for a new flow,
   aggregation mechanisms should be triggered.  Thus, NSIS must have
   interaction with the Mobile IP stack and cannot allow transparent
   operations.






Cheng, et al.            Expires August 5, 2007                 [Page 9]

Internet-Draft              NSIS Multihoming                    Feb 2007


4.2.  Multiple CoAs with single HoA

   This scenario would apply to single interface MNs as well as multi-
   interface MNs.  It is natural that multiple interfaces of a MN would
   obtain different IP address when used simultaneously.  For a single
   interface MN, it is still possible to have multiple IP addresses when
   it is used in a site-multihoming network.  To use the CoAs for the
   same HoA requires some enhancement [8] for MIP.  Details are analyzed
   in the following sub-sections.

   In the static case, i.e. no mobility event, there is no difference in
   the scenario regardless if MN has a single or multiple interfaces.
   However, more complicated mobility scenarios are possible with multi-
   interface MNs.  These will be discussed in section 5.2.

4.2.1.  Using bi-directional tunnel via home agents

   In this scenario, when CN is the QNI, special functionalities should
   be provided by the Home Agents, e.g. the multiplexing of signaling
   messages, indicating binding relationship of the different paths,
   etc.  When MN is the QNI, the Home Agent only needs to merge the
   different signaling messages.  Following example illustrated this
   clearly.

   Figure 3 shows the possible data paths when multiple CoAs are used
   for a single HoA.  Assuming MN obtained two CoAs, CoA-a and CoA-b
   respectively, using them simultaneously requires registering the two
   CoAs with the same Home Agent.  In order to achieve this, the
   Multiple CoA Registration enhancement [8] needs to be supported at
   HA.

   In this scenario, CN is unaware of the multiple addresses involved in
   the communication.  Therefore, special NSIS signaling management
   functions are necessary at the HA.

















Cheng, et al.            Expires August 5, 2007                [Page 10]

Internet-Draft              NSIS Multihoming                    Feb 2007


                +--+
                |CN|
                +--+
                 aa           aaaaa Path from CN to HA
                 aa           aaaaa
                 aa
                +--+          AAAAA Tunnel path from HA
                |HA|                   to CoA-a
                +--+
                 AB
                 AB           BBBBB Tunnel path from HA
                 AB                    to CoA-b
               +----+
       AAAAAAAA|HCRN|BBBBBBBB
       A       +----+       B
       A                    B
       A                    B
     +---+                +---+
     |AR1|                |AR2|
     +---+                +---+
       A                    B
       A                    B
       A        +--+        B
       AAAAAAAAA|MN|BBBBBBBBB
                +--+



   Figure 3.  Multiple CoAs, Single HoA, with Bi-direction tunnel

   When the signaling is initiated by the CN, e.g.  CN being the QNI,
   only one signaling session is created, since to CN, there is only one
   MN address, the HoA.  Therefore, the E2E signaling from CN towards MN
   will carry a SID, say SID-a, and a FlowID derived from MN's HoA, say
   FID-a.  When this E2E signaling message arrives at HA, flow filtering
   rules would apply to the FlowID to decide the path (CoA).

   For example, if the application session makes use of two ports,
   Port-A and Port-B, and the filtering rule at HA may be set such that
   packets for Port-A is to be sent to CoA-a and packets for Port-B is
   to be sent to CoA-b.  Obviously, the FlowID would cover both Port-A
   and Port-B.  Therefore, when applying the filtering rule to the
   FlowID, all matches should be identified, instead of stopping at the
   first match as for processing normal data packet.

   In the above example, HA would identify two CoAs for the FlowID of
   the received E2E signaling message.  This means data flow for the
   session may go via both paths.  The HA would need to create signaling



Cheng, et al.            Expires August 5, 2007                [Page 11]

Internet-Draft              NSIS Multihoming                    Feb 2007


   session for both of the tunnel paths, e.g.  SID-A and SID-B.
   Adjustment of the signaling application data, e.g.  QSPEC, is
   necessary.  However, information for such adjustment might not be
   available at HA, e.g. how many percent of the traffic goes to Port-A,
   and thus CoA-a.  Therefore, collaboration of MN is required, i.e. for
   the tunnel sessions, MN should be the QNI.

   Lastly, when forwarding the E2E signaling message to MN, HA could
   choose to send one copy of the message over any of the path, carrying
   two BOUND_SESSION_IDs set to SID-A and SID-B respectively.
   Alternatively, HA could duplicate the E2E signaling message and send
   one over each of the two paths.  In this case, a special flag is
   necessary to indicate to the MN how to treat the message
   duplications.

   When the signaling is initiated by MN, e.g.  MN being the QNI,
   operations would be subtle.  MN can initiate two flows, using either
   two SID and BOUND_SESSION_ID objects, or using one SID and two
   FlowIDs with REPLACE flag unset.  Normal tunnel operation defined in
   [7] is sufficient for the paths between MN and HA.  The HA however,
   needs to merge the two flows, and presents one single signaling
   towards the CN.

4.2.2.  Using Route Optimization

   In this scenario, current defined NSIS procedure is adequate to deal
   with the multiple paths, e.g. either using multiple SID with
   BOUND_SESSION_ID objects, or using the same SID and unsetting REPLACE
   flag.

   Figure 4 shows the possible route optimized data paths when multiple
   CoAs are used for a single HoA.  In this case, MN uses the two CoA,
   CoA-a and CoA-b, to communicate with CN directly.  The two data paths
   merge at the CN side CRN (CCRN).  Obviously, to support this
   scenario, CN needs to implement certain enhancement of the Mobile IP,
   e.g. the Multiple CoA Registration [8].















Cheng, et al.            Expires August 5, 2007                [Page 12]

Internet-Draft              NSIS Multihoming                    Feb 2007


                +--+
                |CN|
                +--+          aaaaa Path from CN to CoA-a
                 ab
                 ab
                 ab           bbbbb Path from CN to CoA-b
                 ab
               +----+
       aaaaaaaa|CCRN|bbbbbbbb
       a       +----+       b
       a                    b
       a                    b
     +---+                +---+
     |AR1|                |AR2|
     +---+                +---+
       a                    b
       a                    b
       a        +--+        b
       aaaaaaaaa|MN|bbbbbbbbb
                +--+



   Figure 4.  Multiple CoA, Single HoA, with Route Optimization

   In this case, the CN is aware of the two flows used in the transport.
   However, the use of multiple addresses is transparent to applications
   above the IP layer due to the use of Mobile IP.

   As shown in Figure 4, the two paths have different destination
   addresses.  Therefore, two different MRIs are required for the NSIS
   signaling, i.e. separate signaling should be carried out for the two
   paths.  As the two paths are used for the same application level
   session, user might want to create certain association between them,
   e.g. for resource sharing and better signaling state management.
   This could be achieved by using the same Session ID for the two
   flows, or using different Session ID and BOUND_SESSION_ID object in
   the signaling message.

   In case the same Session ID is used, the REPLACE flag should be unset
   in the QoS NSLP message so that the QNEs along the path between CN
   and CCRN could distinguish this from a mobility event.  The RMF on
   those QNEs could decide the proper treatment for the two flows, e.g.
   allow sharing of the resources, etc.  This does not require any
   special functionality for the QoS NSLP signaling and state machine
   management.

   In case different Session IDs are used, at least one of the flow's



Cheng, et al.            Expires August 5, 2007                [Page 13]

Internet-Draft              NSIS Multihoming                    Feb 2007


   signaling messages should carry a BOUND_SESSION_ID object that
   indicates the Flow ID used by the flow.  In this case, the common
   QNEs follow procedures specified in QoS NSLP [4] regarding the
   session binding.

4.3.  Multiple CoAs with Multiple HoAs

   This scenario is basically a combination of the scenarios introduced
   in section 4.1 and 4.2, i.e.  Single CoA with multiple HoAs, and
   Multiple CoAs with single HoA.  Therefore, all the issues discussed
   earlier apply here.  Besides the above issues, new requirements on
   efficient path selection would apply as well.

   The new requirements arise due to the multiple potential paths
   between the CN and MN.  Generally, for a MN of m CoAs and n HoAs, the
   number of possible paths is m*n.  Therefore, picking the right path
   combination for the application session becomes an issue.  Carrying
   end-to-end queries over all the possible paths obviously has high
   overheads and causes long delay in path setup.  Optimization is
   desired for expediting the path selection process.

   When Route optimized path is used in this scenario, the number of
   possible data path is largely determined by the number of CoAs used
   by the MN.  In this case, one CoA could be shared by multiple HoAs.
   Therefore, issues discussed in section 4.1.2 also apply here.


























Cheng, et al.            Expires August 5, 2007                [Page 14]

Internet-Draft              NSIS Multihoming                    Feb 2007


5.  Multihoming Scenarios with mobility

   This section analyzes multihoming scenarios introduced in section 4
   when mobility events are considered.  Same subsection structure as
   section 4 is maintained, so that cross reference could be done
   easily.

5.1.  Single CoA with multiple HoAs

   Static case of this configuration is introduced in section 4.1.  In
   this section, focus is on the mobility aspect of the operation.

5.1.1.  Using bi-direction tunnel via home agents

   Since the MN has only one CoA, change of the location results in
   change of paths from MN to all Home Agents.  However, this change
   should only affect the tunnel section between MN and its Home Agents.
   The end to end section, i.e. from CN to Home Agents, should not be
   altered.  Therefore, the change of the CoA should not trigger updates
   of the path for end to end section.

5.1.2.  Using Route Optimization

   When Route Optimization paths are used for the communication, there
   is essential one address pair in use, as shown in Figure 2.

   When the MN changes CoA (results from a mobility event), both paths
   will change.  Therefore, multiple signaling messages need to be sent
   to update the paths.  Signaling optimization is desired to avoid a
   surge in the signaling, especially when MRIs are identical.  Using
   aggregation would help in this aspect.

5.2.  Multiple CoAs with single HoA

   As discussed in section 4.2, to support this scenario requires
   extension of the Mobile IP stack, e.g. the multi-CoA Registration
   enhancement.

   In this scenario, the multiple CoAs of the MN may change
   independently.  For example, a MN that is handing over to another
   WiFi hotspot may keep the CoA assigned to its WiMAX access.

   Issues arise from this independence in mobility as current NSIS
   message does not indicate such relationships between sessions.
   Details of the issue are analyzed in the following subsections using
   examples.





Cheng, et al.            Expires August 5, 2007                [Page 15]

Internet-Draft              NSIS Multihoming                    Feb 2007


5.2.1.  Using bi-direction tunnel via home agents

   In this scenario, due to the MN mobility and multihoming, different
   types of CRN would be generated.  To distinguish these CRNs is
   essential for the correct signaling of the flows.

   As shown in Figure 5, MN originally has two CoAs, CoA-a and CoA-b,
   used for the communication with CN via the Home Agent (HA).  Assuming
   the HA supports the mCoA extension of Mobile IP, MN registers both
   CoAs with the HA.  CN is unaware of the multiple CoAs used for the
   communication.

   The two tunnel paths from HA to MN split at node HCRN.  Thus, path
   between HA and HCRN are shared by the two flows.  As indicated in
   section 4.2.1, there are two possible ways of signaling the
   relationship of the two flows, using the same SID or different SIDs.

   At certain point of time, MN obtains a new CoA for one of its
   interface, say CoA-c, and registers it with HA to replace an existing
   CoA.  As shown in Figure 5, path from the new CoA-c to the HA (marked
   with 'C') joins the path from CoA-A to HA (marked with 'A') at TCRN,
   and in turn joins the path from CoA-b to HA (marked with 'B') at
   HCRN.

   In case the CoA-c is to replace CoA-a, path marked 'B' and 'C' should
   co-exist, and path marked 'A' should be torn down.  Therefore, the
   signaling message for path 'C' should include information to allow
   TCRN identifying the relationship of the two paths ('C' replace 'A'),
   and at the same time, it should also allow HCRN identifying the other
   relationship ('C' does not replace 'B').

   In case the CoA-c is to replace CoA-b instead, path marked 'A' and
   'C' should co-exist, and path marked 'B' should be torn down.
   Therefore, signaling message for path 'C' should contain information
   for TCRN and HCRN to draw different conclusion as in the above
   example.

   However, the current NSIS signaling message does not contain
   information to that effect.  Detail analysis for cases where the same
   SID is used and different SIDs are used is provided below.











Cheng, et al.            Expires August 5, 2007                [Page 16]

Internet-Draft              NSIS Multihoming                    Feb 2007


                     +---+
                     |CN |
                     +---+
                       a           aaaaa Path from CN to HA
                       a
                       a
                     +---+         AAAAA Tunnel path from HA
                     |HA |                  to CoA-a
                     +---+
                      ABC
                      ABC          BBBBB Tunnel path from HA
                      ABC                   to CoA-b
                    +----+
            AAAAAAAA|HCRN|BBBBBBBB
            ACCCCCCC+----+       B BBBBB Tunnel path from HA
            AC                   B          to CoA-c
          +----+                 B
      AAAA|TCRN|CCCC             B
      A   +----+   C             B
      A            C             B
    +---+        +---+         +---+
    |AR1|        |AR3|         |AR2|
    +---+        +---+         +---+
      A            C             B
      A          +---+           B
      AAAAAAAAAAA|MN |BBBBBBBBBBBB
                 +---+



   Figure 5.  Mobility scenario for MN with multi-CoA and single HoA

   When the same SID is used for all the tunneled paths between HA and
   MN, special techniques are required.  For example, the REPLACE flag
   in the QoS NSLP message for path 'A' and 'B' should be unset, so that
   QNEs on the common path do not initiate tear-down of the other path.
   Problem arises when the signaling message for path 'C' comes in.  If
   the REPLACE flag is also unset in this message, those QNEs would not
   know path 'C' is replacing one of the paths.  However, if the REPLACE
   flag is set, it might cause QNE to replace state for both path 'A'
   and 'B' with that for path 'C'.  A more detailed analysis of this
   case is available in Mobility Applicability draft section 5.2.5 [5].
   Solving this problem might require some extra identifier to indicate
   the relationships.  The BID introduced in [8] would be a good
   candidate for such use.

   When different SIDs are used for tunneled paths, signaling message
   for path 'C' should take the same SID as that of path 'A' if CoA-c is



Cheng, et al.            Expires August 5, 2007                [Page 17]

Internet-Draft              NSIS Multihoming                    Feb 2007


   replacing CoA-a.  (And, it should take the same SID as that of path
   'B' if CoA-c is replacing CoA-b.)  The REPLACE flag of the QoS NSLP
   message should be set.  Therefore, QNEs on the common path would
   properly update the QoS states.

   Session Binding might pose a problem for the different SIDs case.
   When different SIDs are used for path 'A' and 'B', they should be
   bound using BOUND_SESSION_ID object.  Therefore, when path 'C'
   replaces one of the paths, say path 'A', it shall not cause the
   removal of the other path, e.g. path 'B'.  Therefore, the
   BOUND_SESSION_ID also needs to have certain type indication for this
   case.

5.2.2.  Using Route optimization

   In mobility aspect, this case is quite similar to that described in
   section 5.2.1.  Replacing the HA with CN in Figure 5, it could
   represent the possible data path of this case.  Therefore, the
   problems discussed above would generally apply to this case also.

5.3.  Multiple CoAs with a single HoA

   This is a complicated scenario that encompasses all possible
   combinations of cases described in section 5.1 and 5.2.  Therefore,
   issues discussed in these sections would also apply here.


























Cheng, et al.            Expires August 5, 2007                [Page 18]

Internet-Draft              NSIS Multihoming                    Feb 2007


6.  Problem statement for NSIS multihoming support

   As analyzed in section 4 and 5, current NSIS signaling has some
   glitches when dealing with the multihoming systems.  A summary of the
   functionalities that is lacking or needs to be provided by NSIS is
   provided below:

   - Reflecting session binding relationship when tunneling is used.  As
   discussed in section 4.1.1, when bounded session passing through
   tunnels separately, the created tunnel sessions should also be bound.

   - Interaction with Mobile IP stack should trigger aggregation
   mechanism when same MRI is resulted for different sessions.  As
   discussed in section 4.1.2, it is possible that multiple sessions
   have the same MRI when Route Optimization is used.  Session
   aggregation at end host would be necessary since intermediary QNEs
   has no way to separate the flows.

   - Signaling multiplexing at Home Agents when multihoming is supports.
   As discussed in section 4.2.1, CN could be unaware of the multiple
   CoA address used by MN.  Therefore, only one session would be
   initiated by CN, and HA needs to perform multiplexing accordingly.

   - Expedited path query and selection.  As discussed in section 4.3,
   there could be large number of potential paths used for the
   application session.  In order to achieve faster transition and less
   service interruption, faster path query would be desired.

   - Distinguish path relationships in mobile environment.  As discussed
   in section 5.2.1, signaling message should contain information to
   indicate relationship with the existing paths (replace or coexist) in
   the multi-path environment.

   - Indicating the type of binding for the multihoming support.  As
   discussed in section 5.2.1, such indicating would be necessary to
   avoid accidental removal of the parallel sessions.















Cheng, et al.            Expires August 5, 2007                [Page 19]

Internet-Draft              NSIS Multihoming                    Feb 2007


7.  Security Considerations

   Generally, the signaling security is covered with the current NSIS
   protocol design [7][4].  However, when work in the multihoming
   environment, a few other security issue arises.  For example,
   multiple flows from different IP address are bound and share the
   resources in multihoming scenarios.  This would require authorization
   schemes that does not relying on IP address as the user identity.
   Other than that, some of the intermediary nodes, e.g. the Home Agent,
   may have extended access to the e2e signaling message.  This would
   require new security measures to counter possible attacks.

   These security issues would be considered in future version of the
   draft.





































Cheng, et al.            Expires August 5, 2007                [Page 20]

Internet-Draft              NSIS Multihoming                    Feb 2007


8.  Conclusion

   This draft provides a scenario by scenario analysis of the NSIS
   applicability in multihoming environment.  Operation issues were
   identified in the discussion.  Causes of the issues and relevant
   requirements on NSIS are summarized at the end of the draft.
   Studying those functionalities identified as lacking in the current
   NSIS would provide directions for extending NSIS's coverage for an
   emerging area.










































Cheng, et al.            Expires August 5, 2007                [Page 21]

Internet-Draft              NSIS Multihoming                    Feb 2007


9.  Acknowledgements

   This section contains the acknowledgements.
















































Cheng, et al.            Expires August 5, 2007                [Page 22]

Internet-Draft              NSIS Multihoming                    Feb 2007


10.  References

10.1.  Normative Reference

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

   [2]  IETF, "Mobile Nodes and Multiple Interfaces in IPv6 (monami6)",
        WG Charter
         http://www.ietf.org/html.charters/monami6-charter.html,
        Dec 2006.

   [3]  IETF, "Site Multihoming by IPv6 Intermediation (shim6)", WG
        Charter  http://www.ietf.org/html.charters/shim6-charter.html,
        May 2006.

   [4]  Manner, J., "NSLP for Quality-of-Service Signaling", Internet
        Draft:  draft-ietf-nsis-qos-nslp-12, Work in progress ,
        Oct 2006.

   [5]  Lee, S., "Applicability Statement of NSIS Protocols in Mobile
        Environments", Internet Draft:
         draft-ietf-nsis-applicability-mobility-signaling-05, Work in
        progress , June 2006.

   [6]  Abley, J., Black, B., and V. Gill, "Goals for IPv6 Site-
        multihoming Architectures", BCP 14, RFC 3582, August 2003.

   [7]  Shen, C., "NSIS Operation Over IP Tunnels", Internet Draft:
         draft-ietf-nsis-tunnel-01, Work in Progress , Oct 2006.

   [8]  Wakikawa, R., "Multiple Care-of-Address Registration", Internet
        Draft:  draft-ietf-monami6-multiplecoa-01, Work in progress ,
        Oct 2006.

10.2.  Informative References

   [9]  Montavont, N., "Analysis of Multihoming in Mobile IPv6",
        Internet Draft:  draft-ietf-monami6-analysis-01, Work in
        progress , Oct 2006.











Cheng, et al.            Expires August 5, 2007                [Page 23]

Internet-Draft              NSIS Multihoming                    Feb 2007


Authors' Addresses

   Hong Cheng
   Panasonic Singapore Laboratories
   Block 1022, Tai Seng Industrial Estate
   #06-3530, Tai Seng Avenue
   Singapore  534415
   Singapore

   Phone: +65 6550 5477
   Email: hong.cheng@sg.panasonic.com


   Takako Sanda
   Matsushita Electric Industrial Co., Ltd. (Panasonic)
   5-3, Hikarino-oka, Yokosuka City
   Kanagawa  239-0847
   Japan

   Phone: +81 50 3687 6563
   Email: sanda.takako@jp.panasonic.com


   Toyoki Ue
   Matsushita Electric Industrial Co., Ltd. (Panasonic)
   5-3, Hikarino-oka, Yokosuka City
   Kanagawa  239-0847
   Japan

   Phone: +81 50 3687 6562
   Email: ue.toyoki@jp.panasonic.com


   Shivanajay Marwaha
   Panasonic Singapore Laboratories
   Block 1022, Tai Seng Industrial Estate
   #06-3530, Tai Seng Avenue
   Singapore  534415
   Singapore

   Phone: +65 6550 5414
   Email: shivanaja.marwaha@sg.panasonic.com









Cheng, et al.            Expires August 5, 2007                [Page 24]

Internet-Draft              NSIS Multihoming                    Feb 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.


Acknowledgment

   Funding for the RFC Editor function is provided by the IETF
   Administrative Support Activity (IASA).





Cheng, et al.            Expires August 5, 2007                [Page 25]