One document matched: draft-shen-nsis-tunnel-01.txt

Differences from draft-shen-nsis-tunnel-00.txt





IETF Next Steps in Signaling                                     C. Shen
Internet-Draft                                            H. Schulzrinne
Expires: April 27, 2006                                      Columbia U.
                                                                  S. Lee
                                                                 J. Bang
                                                             Samsung AIT
                                                        October 24, 2005


                     NSIS Operation Over IP Tunnels
                     draft-shen-nsis-tunnel-01.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 April 27, 2006.

Copyright Notice

   Copyright (C) The Internet Society (2005).

Abstract

   This draft presents an NSIS operation over IP tunnels scheme using
   QoS NSLP as the NSIS signaling application.  Both sender-initiated
   and receiver-initiated reservation modes are discussed.  The scheme
   proposes a separate signaling session inside the tunnel.  Packets
   belonging to qualified tunnel sessions are assigned special flow IDs



Shen, et al.             Expires April 27, 2006                 [Page 1]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   to be distinguished from the rest of the tunnel traffic.  The end-to-
   end session and its corresponding tunnel session are associated with
   each other when necessary; so that adjustment in one session may be
   reflected in the other.


Table of Contents

   1.  Requirements notation  . . . . . . . . . . . . . . . . . . . .  4
   2.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  4
     2.1.  IP Tunneling Mechanisms  . . . . . . . . . . . . . . . . .  4
     2.2.  Different Signaling Capabilities of IP Tunnels . . . . . .  5
   3.  Overall Protocol Design  . . . . . . . . . . . . . . . . . . .  6
   4.  Protocol Design Details  . . . . . . . . . . . . . . . . . . .  7
     4.1.  Packet Classification Over the Tunnel  . . . . . . . . . .  7
     4.2.  Tunnel Signaling and its Association with End-to-End
           Signaling  . . . . . . . . . . . . . . . . . . . . . . . .  8
   5.  Protocol Operation for Individual Tunnel Signaling . . . . . . 11
     5.1.  Basic Sender-Initiated Signaling over IP Tunnels . . . . . 11
     5.2.  Basic Receiver-Initiated Signaling over IP Tunnels . . . . 12
   6.  Protocol Operation for Aggregate and Mixed Signaling
       Tunnels  . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
     6.1.  Tunnel With Only One Aggregate Session . . . . . . . . . . 14
     6.2.  Tunnel With Multiple Aggregate Sessions  . . . . . . . . . 14
     6.3.  Adjustment of Configured Tunnel Sessions . . . . . . . . . 14
     6.4.  Protocol Operation for Mixed Signaling Tunnels . . . . . . 15
   7.  Message Processing Rules for Selected End-to-End QoS NSLP
       Messages at Tunnel Endpoints . . . . . . . . . . . . . . . . . 15
     7.1.  End-to-End QUERY Message at Tentry . . . . . . . . . . . . 15
     7.2.  End-to-End QUERY Message at Texit  . . . . . . . . . . . . 16
     7.3.  End-to-End RESERVE Message at Tentry . . . . . . . . . . . 16
     7.4.  End-to-End RESERVE Message at Texit  . . . . . . . . . . . 18
     7.5.  Special Processing Rules for Many-to-One Mapping
           Tunnels  . . . . . . . . . . . . . . . . . . . . . . . . . 19
   8.  Other Considerations . . . . . . . . . . . . . . . . . . . . . 19
     8.1.  Other Types of NSLP  . . . . . . . . . . . . . . . . . . . 19
     8.2.  IPSEC Flows  . . . . . . . . . . . . . . . . . . . . . . . 20
     8.3.  NSIS-Tunnel and Mobility . . . . . . . . . . . . . . . . . 20
   9.  Security Considerations  . . . . . . . . . . . . . . . . . . . 21
   10. Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
     10.1. Summary of RSVP Operation Over IP Tunnels  . . . . . . . . 21
     10.2. Various Design Alternatives  . . . . . . . . . . . . . . . 21
       10.2.1.  Carrying Signaling Messages over the Tunnel . . . . . 21
       10.2.2.  Packet Classification over the Tunnel . . . . . . . . 22
       10.2.3.  Tunnel Binding Methods  . . . . . . . . . . . . . . . 22
       10.2.4.  Tunnel Binding Indication . . . . . . . . . . . . . . 23
       10.2.5.  Carrying the Tunnel Binding Object  . . . . . . . . . 23
       10.2.6.  Alternative Ways of End-to-End and Tunnel Session



Shen, et al.             Expires April 27, 2006                 [Page 2]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


                Interaction . . . . . . . . . . . . . . . . . . . . . 23
   11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 24
   12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 24
     12.1. Normative References . . . . . . . . . . . . . . . . . . . 24
     12.2. Informative References . . . . . . . . . . . . . . . . . . 24
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 27
   Intellectual Property and Copyright Statements . . . . . . . . . . 28












































Shen, et al.             Expires April 27, 2006                 [Page 3]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


1.  Requirements notation

   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 [1].


2.  Introduction

   IP tunnel mechanisms are widely used in the Internet for various
   purposes.  When a tunnel is used to transfer signaling messages, e.g.
   NSIS messages, the signaling messages themselves usually become
   invisible inside the tunnel.  In other words, the tunnel behaves as a
   logical link that does not support signaling in the end-to-end path.
   If end-to-end NSIS signaling support is desired for a path containing
   tunnels, it is necessary to define a scheme that allows NSIS
   operation over IP tunnels.  This draft describes such a scheme.  We
   assume QoS NSLP as the NSIS signaling application.

2.1.  IP Tunneling Mechanisms

   There are a number of common tunneling mechanisms used in the
   Internet.  A non-exhausted list of them is as follows,

   o  Generic Routing Encapsulation (GRE) [4] is a mechanism for
      encapsulating arbitrary packets within an arbitrary transport
      protocol.  Generic Routing Encapsulation over IPv4 Networks
      (GREIP4) [5] addresses the case of using IPv4 as the delivery
      protocol or the payload protocol and the special case of IPv4 as
      both the delivery and payload.  Generic Routing Encapsulation
      (GREIP4A) [17] presented a modified version of [4], in particular,
      some flag bits in the original specification have been deprecated.
   o  IP Encapsulation within IP (IP4INIP4) [7] is a method of tunneling
      IPv4 packets using an additional IPv4 header.  Minimal
      Encapsulation within IP (MINENC) [8] describes a way to reduce the
      size of the "inner" IP header used in [7] when the original
      datagram is not fragmented.
   o  Generic Packet Tunneling in IPv6 Specification (IP6GEN) [11]
      specifies a method by which a packet is carried as payload within
      an IPv6 packet by being encapsulated in an IPv6 header, and
      optionally, a set of IPv6 extension headers.
   o  IPv6 over IPv4 tunneling (IP6INIP4) [6] encapsulates IPv6 packets
      within IPv4 headers to carry them over IPv4 routing
      infrastructures.
   o  IPSEC [9] has a tunnel mode with the use of Encapsulating Security
      Payload (ESP) [10].  The tunneled IP packets are encrypted and the
      ESP is placed before the encapsulated IP header.




Shen, et al.             Expires April 27, 2006                 [Page 4]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   The above tunneling mechanisms fall into two broad categories
   according to the encapsulating (delivery) header format:

   1.  Normal IP in IP Encapsulation: the encapsulating header is just a
       standard IP header.  This group includes IP4INIP4, IP6INIP4,
       IP6GEN.
   2.  Modified IP in IP Encapsulation: the encapsulating header is a
       standard IP header plus additional information.  This group
       includes all GRE-related IP tunneling, MINENC and IPSEC tunneling
       mode.  The additional information in these cases is the GRE
       header, minimum encapsulation header and ESP header respectively.
       This information is usually placed between the encapsulating IP
       header and the original IP header.  (MINENC is an exception
       because it modifies the original IP header).  Note that in the
       IPSEC case, the original IP header is also encrypted along with
       the original IP payload.

2.2.  Different Signaling Capabilities of IP Tunnels

   By default any end-to-end signaling messages arriving at the tunnel
   endpoint will be encapsulated the same way as data packets.  Tunnel
   intermediate nodes do not identify them as signaling messages.
   Therefore the tunnel appears as a signaling unaware logical link to
   the end-to-end session.

   A signaling aware tunnel may participate in a signaling network in
   various ways.  For example, RFC 2746 [18] identifies two types of QoS
   aware tunnels: a tunnel that can promise that some overall level of
   resources is available to carry traffic, but not to allocate
   resources specifically to individual data flows; or a tunnel that can
   make reservations for individual end-to-end data flows.

   We call these two types of tunnels aggregate signaling tunnel and
   individual signaling tunnel respectively.  The key feature that
   distinguishes individual signaling tunnels from aggregate signaling
   tunnels is that in individual signaling tunnels new tunnel sessions
   are created and torn down dynamically as end-to-end sessions come and
   go.  Note that the tunnel sessions in aggregate signaling tunnels
   could be a configured tunnel that never gets changed, or could be an
   aggregate tunnel session that is dynamically adjusted as the actually
   used session resources increase or decrease.  On the other hand,
   individual signaling tunnels may also contain multiple tunnel
   sessions for the same application, e.g. an audio and its associated
   video stream.

   A tunnel may also be a mixed one that combines the properties of the
   aggregate signaling tunnel and individual signaling tunnel.




Shen, et al.             Expires April 27, 2006                 [Page 5]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


3.  Overall Protocol Design

   This document presents a scheme to enable NSIS operation over IP
   tunnels with different tunnel capabilities.  The design goals of the
   scheme are as follows,

   o  For best effort tunnel, make sure NSIS messages traverse the link
      correctly, and the presence of the non-NSIS aware link is
      detected.
   o  For aggregate and individual signaling tunnels, make sure proper
      signaling is carried out in the tunnel for the end-to-end flow.
   o  Work with most, if not all, existing IP in IP tunneling schemes.
   o  Place the additional tunnel related functionalities only in one or
      both of the tunnel endpoints.
   o  If possible, make NSIS tunnel signaling handle specific events in
      a consistent way as that of NSIS signaling without tunneling.  An
      example is mobility interaction.

   The overall design of NSIS operation over IP tunnels is conceptually
   similar to RSVP operation over IP tunnels [18].  (A short summary of
   [18] is provided in appendix Section 10.1).  However, the scheme
   described in this document also addresses the important differences
   of NSIS from RSVP, e.g.,

   o  NSIS is based on a two-layer architecture, namely a signaling
      transport layer and a signaling application layer.  It is designed
      as a generic framework to accommodate various signaling
      application needs.  The basic RSVP protocol does not have a layer
      split and is only for QoS signaling.
   o  NSIS QoS NSLP allows both sender-initiated and receiver-initiated
      reservations; RSVP only supports receiver-initiated reservations.
   o  NSIS deals only with unicast; RSVP also supports multicast.
   o  NSIS integrates new features, such as the Session ID, to
      facilitate operation in specific environments (e.g. mobility and
      multi-homing).

   From a high level point of view, the main issues in a signaling
   operation over IP tunnel scheme are, how packet classification is
   performed inside the tunnel, and how signaling is carried out inside
   the tunnel.

   Packets belonging to qualified data flows need to be recognized by
   tunnel intermediate nodes to receive special treatment.  Packet
   classification is traditionally based on flow ID, which is derived
   from various fields in Message Routing Information (MRI).  The
   problem is, after a typical IP-in-IP tunnel encapsulation, all
   packets going through the tunnel appear as having the same flow ID
   (which consists of the Tunnel Entry (Tentry) address and Tunnel Exit



Shen, et al.             Expires April 27, 2006                 [Page 6]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   (Texit) address.  Therefore the flow ID for signaled flows needs to
   contain further demultiplexing fields in order to be distinguished
   from non-signaled flows, and also from one another among all signaled
   flows.

   The special flow ID for signaled flows inside the tunnel then needs
   to be carried in tunnel signaling messages to set up or modify the
   state information in tunnel intermediate nodes.  The original end-to-
   end signaling messages do not contain tunnel specific parameters such
   as the tunnel flow ID and tunnel adjusted QoS parameters.  Therefore,
   separate tunnel signaling sessions are generated and maintained
   between the tunnel endpoints, as in the case of RSVP operation over
   IP tunnels [18].  When end-to-end signaling sessions and tunnel
   signaling sessions are carried out separately, it will be necessary
   in many cases to maintain the state association between the end-to-
   end session and its corresponding tunnel session so that any change
   to one session may be reflected in the other.

   In the next section, we will illustrate details on packet
   classification over the tunnel, signaling over the tunnel as well as
   association of end-to-end and tunnel signaling.


4.  Protocol Design Details

4.1.  Packet Classification Over the Tunnel

   Tunnel flows need to be assigned special flow IDs in order to allow
   tunnel packet classification.  A flow can be an individual flow or an
   aggregate flow.  Possible flow ID formats that may be used to
   identify individual tunnel flows are listed below:

   o  Selected fields from the base IP header of the tunnel encapsulated
      packet (outer IP header).  For example, the IP source and
      destination address fields, which contain the IP addresses of
      Tentry and Texit, together with another field for tunnel-wide
      demultiplexing.  This could be the IPv6 flow label field, or the
      Traffic Class (TC) field.  Note that the TC field can also be used
      in DiffServ to carry DiffServe Code Point (DSCP) and thus
      represent an aggregate flow.  As long as individual flow
      classification is processed before aggregate flow classification,
      or a longest match kind of packet classifier is used, the tunnel
      flow demultiplexing with TC field should work.  In the rare cases
      where these conditions could not be satisfied, it is still
      possible to choose different range of DSCP values so that the
      values used for individual tunnel flow demultiplexing do not
      collide with those used for DiffServ aggregate flows.  Compared to
      the IPv6 flow label approach, the tunnel flow ID containing DSCP



Shen, et al.             Expires April 27, 2006                 [Page 7]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


      can be applied to both IPv4 and IPv6 and is probably easier to
      deploy.  Its drawback is that the small number of bits in the DSCP
      field limits the total number of individual flows that can be
      distinguished in the tunnel.  Overall, these flow ID formats in
      this group enable efficient packet classification over the tunnel
      without introducing additional processing requirements on the
      existing infrastructure.  They are also easy to deploy.

   o  Selected fields from the tunnel base IP header plus additional
      information outside the base IP header but still in the tunnel
      encapsulation header.  This applies to modified IP-in-IP
      encapsulation as we defined in Section 2.1.  An example of this
      additioanl information is the SPI field for IPSEC tunnels.
      Comparing with the first group, the flow ID formats in this group
      poses more requirements at the NSIS protocol side because it uses
      information unique to the specific tunnel mechanism.  NSIS thus
      needs to be specifically tuned to recognize that information as
      part of a signaling message.  This is similar to how RFC 2207 [19]
      has extended RSVP to accommodate IPSEC.

   o  UDP header insertion.  Inserting a new UDP header between the
      tunnel IP header and the tunnel payload provides additional
      demultiplexing information for a tunnel flow.  The drawback of the
      flow ID format in this group, as compared to the above two, is the
      additional UDP header overhead both for bandwidth and processing.
      In addition, this approach modifies the basic tunneling mechanism
      at the Tentry, so Texit will also need to be aware of the special
      encapsulation in order to correctly decapsulate and forward
      packets further along the path.


   Most of the above flow ID formats may also be used for aggregate
   tunnel flows.  For example, a common aggregate flow ID contains the
   addresses of tunnel endpoints and the DSCP value.  When additional
   interfaces at the tunnel endpoints are available, these addresses may
   also be used to form aggregate flow ID.  For example, the IP address
   of an additional interface for a Tentry plus the IP address of the
   Texit, constitute an aggregate flow ID.

   The choice of using which of the above flow ID format is left to a
   policy mechanism outside the scope of this document.  Tunnel
   signaling is performed based on the chosen flow ID and Tentry should
   encapsulate all incoming packets for the specific data flows
   according to the chosen flow ID format.

4.2.  Tunnel Signaling and its Association with End-to-End Signaling

   Tunnel signaling messages contain tunnel specific parameters such as



Shen, et al.             Expires April 27, 2006                 [Page 8]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   tunnel MRI and tunnel adjusted QoS parameters.  But the formats of
   tunnel signaling messages are the same as end-to-end signaling
   messages and tunnel signaling is carried out according to the same
   signaling flows of the end-to-end signaling.  The main challenge is
   therefore the interaction between tunnel signaling and end-to-end
   signaling.  The interaction is achieved by special functionalities
   supported in the NSIS-aware tunnel endpoints.  These special
   functionalities include assigning special tunnel flow IDs, creating
   tunnel session association, notifying the other endpoint about tunnel
   association, adjusting one session based on change of the other
   session, encapsulating (decapsulating) packets according to the
   chosen tunnel flow ID at Tentry (Texit), etc.  In most cases, we
   expect to have bi-directional tunnels, where both endpoints are NSIS-
   tunnel aware.

   When both Tentry and Texit need to be NSIS-tunnel aware, it is
   necessary for one NSIS-tunnel aware endpoint to learn whether the
   other endpoint also has the same capability.  This document assumes
   that the NSIS-tunnel awareness is optional and only needs to be
   deployed at tunnel endpoints.  Therefore a tunnel capability
   discovery mechanism will be needed.  This mechanism is still
   considered an open issue.  It may be defined at the GIST layer or at
   the NSLP layer.  It may be defined as a general NSIS-tunnel aware
   indication (for various NSLPs) or it may indicate specifically with
   which NSLPs the tunnel operation is supported.  At the GIST layer,
   the NSIS-tunnel handling characteristics may be carried in an object
   similar to Network-Layer-Information.  At the NSLP layer, a new
   NODE_CHAR object similar to that in [18] may be defined and exchanged
   between the Tentry and Texit.  The NSIS-tunnel messaging flows shown
   in this document assumes both tunnel endpoints already know the other
   endpoint is also NSIS-tunnel aware, in other words, the NSIS-Tunnel
   capability discovery has been performed already.

   When both Tentry and Texit need to be NSIS-tunnel aware, the endpoint
   that creates the tunnel session needs to notify the other endpoint of
   the association between the end-to-end and tunnel session.  We choose
   to achieve this by using a modified QoS NSLP BOUND_SESSION_ID object.
   This object contains the tunnel session SID that should be bound to
   the SID of the message carrying this object.  This modified object is
   carried in an end-to-end signaling messages and sent between the
   tunnel endpoints like any other tunneled packets inside the tunnel,
   so this object will only be recognized and processed at the tunnel
   endpoint.  NSIS-tunnel aware endpoints recognize the object, perform
   tunnel related procedures and then remove the object.

   The QoS NSLP BOUND_SESSION_ID object is used to indicate the
   dependency of two sessions, its format as currently defined does not
   specify the nature of this dependency.  For example, it may be used



Shen, et al.             Expires April 27, 2006                 [Page 9]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   for bi-directional flows or flow aggregation as well.  According to
   [3], when receiving a message with a BOUND_SESSION_ID object, a QNE
   MUST copy the BOUND_SESSION_ID object into all messages it sends for
   the same session.  This is not the desired behavior for its use in
   the context of this document.  For NSIS-tunnel operation, the
   dependency is maintained by the tunnel endpoints and should not be
   propagated further outside the tunnel.  Therefore we propose to add
   to the existing BOUND_SESSION_ID a Binding_Code field as follows.

   Type: BOUND_SESSION_ID

   Length: Fixed - 5 32-bit words

   Value: contains an 8-bit Binding_Code that indicates the nature of
   binding.  The rest specifies the Session ID of the session that must
   be bound to the session associated with the message carrying this
   object.

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                  RESERVED                     |  Binding_Code |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                          Session ID                           +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+



   Figure 1: BOUND_SESSION_ID Object

   Currently defined Binding_Codes are:

   o  0x01 - Tunnel and end-to-end sessions
   o  0x02 - Bi-directional sessions
   o  0x03 - Aggregate sessions
   o  0x04 - Dependent sessions (one session is alive only if the other
      session is also alive)

   More binding codes maybe defined based on the above four atomic
   binding actions.  It is not clear whether we need to allow more than
   one SESSION ID (SID) in the binding object.

   We refer to a BOUND_SESSION_ID object carrying the code 0x01 as a
   tunnel BOUND_SESSION_ID object in this document.



Shen, et al.             Expires April 27, 2006                [Page 10]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


5.  Protocol Operation for Individual Tunnel Signaling

   In an individual signaling tunnel a tunnel session is dynamically
   created and one-to-one mapped to an end-to-end session.  NSIS QoS
   NSLP allows both sender-initiated and receiver-initiated reservation
   modes.  It is possible that the end-to-end and tunnel session may use
   the same or different reservation modes.  Therefore we have the
   following four signaling scenarios:

   o  A. End-to-end session is sender-initiated; tunnel session is
      sender-initiated.
   o  B. End-to-end session is receiver-initiated; tunnel session is
      receiver-initiated.
   o  C. End-to-end session is sender-initiated; tunnel session is
      receiver-initiated.
   o  D. End-to-end session is receiver-initiated; tunnel session is
      sender-initiated.

   We discuss details of scenario A and B in this document.  The other
   two scenarios will be covered in a later version of this draft.

5.1.  Basic Sender-Initiated Signaling over IP Tunnels


     Sender    Tentry      Tnode      Texit     Receiver

       |          |          |          |          |
       | RESERVE  |          |          |          |
       +--------->|          |          |          |
       |          | RESERVE' |          |          |
       |          +=========>|          |          |
       |          |          | RESERVE' |          |
       |          |          +=========>|          |
       |          |       RESERVE       |          |
       |          +-------------------->|          |
       |          |          | RESPONSE'| RESERVE  |
       |          |          |<=========+--------->|
       |          | RESPONSE'|          |          |
       |          |<=========+          |          |
       |          |          |          | RESPONSE |
       |          |          |          |<---------+
       |          |       RESPONSE      |          |
       |          |<--------------------+          |
       | RESPONSE |          |          |          |
       |<---------+          |          |          |
       |          |          |          |          |
       |          |          |          |          |




Shen, et al.             Expires April 27, 2006                [Page 11]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   Figure 2: Sender-Initiated QoS NSLP over IP Tunnel

   This scenario assumes both end-to-end and tunnel sessions are sender-
   initiated.  Figure 2 shows the messaging flow of NSIS operation over
   IP tunnels in this case.  Tunnel signaling messages are distinguished
   from end-to-end messages by a "'" after the message name.  The sender
   first sends an end-to-end RESERVE message which arrives at Tentry.
   If Tentry supports tunnel signaling and determines that an individual
   tunnel session needs to be established for the end-to-end session, it
   chooses the tunnel flow ID, creates the tunnel session and associates
   the end-to-end session with the tunnel session.  It then sends a
   tunnel RESERVE' message matching the requests of the end-to-end
   session toward the Texit to reserve tunnel resources.  Tentry also
   appends to the original RESERVE message with a tunnel
   BOUND_SESSION_ID object containing the SID of the tunnel session and
   sends it toward Texit using normal tunnel encapsulation.

   The tunnel RESERVE' message is processed hop by hop inside the tunnel
   for the flow identified by the chosen tunnel flow ID.  When Texit
   receives the tunnel RESERVE' message, reservation state for the
   tunnel session will be created.  Texit may also send a tunnel
   RESPONSE' message to Tentry.  On the other hand, the end-to-end
   RESERVE message passes through the tunnel intermediate nodes just
   like any other tunneled packets.  When Texit receives the end-to-end
   RESERVE message, it notices the binding of a tunnel session and
   checks the state for the tunnel session.  When the tunnel session
   state is available, it updates the end-to-end reservation state using
   the tunnel session state, removes the tunnel BOUND_SESSION_ID object
   and forwards the end-to-end RESERVE message further along the path
   towards the receiver.  When the end-to-end reservation finishes, an
   end-to-end RESPONSE may be sent back from the receiver to the sender.

5.2.  Basic Receiver-Initiated Signaling over IP Tunnels


     Sender    Tentry      Tnode      Texit     Receiver

       |          |          |          |          |
       |  QUERY   |          |          |          |
       +--------->|          |          |          |
       |          |        QUERY        |          |
       |          +-------------------->|          |
       |          |  QUERY'  |          |          |
       |          +=========>|          |          |
       |          |          |  QUERY'  |          |
       |          |          +=========>|          |
       |          |          |          |  QUERY   |
       |          |          |          +--------->|



Shen, et al.             Expires April 27, 2006                [Page 12]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


       |          |          |          | RESERVE  |
       |          |          |          |<---------+
       |          |          | RESERVE' |          |
       |          |          |<=========+          |
       |          | RESERVE' |          |          |
       |          |<=========+          |          |
       |  RESERVE | RESPONSE'|          |          |
       |<---------+=========>|          |          |
       |          |          | RESPONSE'|          |
       |          |          +=========>|          |
       | RESPONSE |          |          |          |
       +--------->|          |          |          |
       |          |       RESPONSE      |          |
       |          +-------------------->|          |
       |          |          |          | RESPONSE |
       |          |          |          +--------->|
       |          |          |          |          |
       |          |          |          |          |


   Figure 3: Receiver-Initiated QoS NSLP over IP Tunnel

   This scenario assumes both end-to-end and tunnel sessions are
   receiver-initiated.  Figure 3 shows the messaging flow of NSIS
   operation over IP tunnels in this case.  When Tentry receives the
   first end-to-end QUERY message from the sender, it chooses the tunnel
   flow ID, creates the tunnel session and sends a tunnel QUERY' message
   matching the requests of the end-to-end session toward the Texit.
   Tentry also appends to the original QUERY message with a tunnel
   BOUND_SESSION_ID object containing the SID of the tunnel session and
   sends it toward the Texit using normal tunnel encapsulation.

   The tunnel QUERY' message is processed hop by hop inside the tunnel
   for the flow identified by the chosen tunnel flow ID.  When Texit
   receives the tunnel QUERY' message, it should create a reservation
   state for the tunnel session, but it should not send out a tunnel
   RESERVE' message immediately.  This conditional reservation
   processing does not seem to be covered by the current QoS NSLP draft
   [3], so an additional message-specific flag bit in the common header
   of QUERY message may be needed.

   The end-to-end QUERY message passes along tunnel intermediate nodes
   just like any other tunneled packets.  When Texit receives the end-
   to-end QUERY message, it notices the binding of a tunnel session and
   checks state for the tunnel session.  When the tunnel session state
   is available, Texit updates the end-to-end QUERY message using the
   tunnel session state, removes the tunnel BOUND_SESSION_ID object and
   forwards the end-to-end QUERY message further along the path.



Shen, et al.             Expires April 27, 2006                [Page 13]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   When Texit receives the first end-to-end RESERVE message issued by
   the receiver, it finds the reservation state of the tunnel session
   and triggers a tunnel RESERVE' message for that session.  Meanwhile
   the end-to-end RESERVE message will be appended with a tunnel
   BOUND_SESSION_ID object and forwarded towards Tentry.  When Tentry
   receives the tunnel RESERVE', it creates the reservation state for
   the tunnel session and may send a tunnel RESPONSE' back to Texit.
   When Tentry receives the end-to-end RESERVE, it creates the end-to-
   end reservation state and updates it with information from the
   associated tunnel session reservation state.  Then Tentry further
   forwards the end-to-end RESERVE upstream toward the sender.


6.  Protocol Operation for Aggregate and Mixed Signaling Tunnels

   An aggregate signaling tunnel may contain one or more aggregate
   tunnel sessions configured through management interfaces.

6.1.  Tunnel With Only One Aggregate Session

   If only one aggregate session is configured in the tunnel and all
   traffic will receive the reserved tunnel resources, all the packets
   just need to be normal IP-in-IP encapsulated.  If there is only one
   aggregate session configured in the tunnel and only some traffic
   should receive the reserved resources through that aggregate tunnel
   session, then the aggregate tunnel session should be assigned an
   appropriate flow ID.  Qualified packets need to be encapsulated with
   this flow ID.  The rest of the traffic will be normal IP-in-IP
   encapsulated.

6.2.  Tunnel With Multiple Aggregate Sessions

   If there are multiple configured aggregate sessions over a tunnel set
   up by the management interface, these sessions must be distinguished
   by their aggregate tunnel flow IDs based on appropriate flow ID.  In
   this case it is necessary to explicitly bind the end-to-end sessions
   with the specific tunnel sessions.  This binding is provided by the
   tunnel BOUND_SESSION_ID object which is carried in the end-to-end
   signaling message.  Once the binding has been established, Tentry
   should encapsulate qualified data packets from different flows
   according to the associated aggregate tunnel flow ID.  Intermediate
   nodes in the tunnel will then be able to filter these packets to
   receive reserved resources.

6.3.  Adjustment of Configured Tunnel Sessions

   The reservation of a configured tunnel session may or may not be
   adjustable.  When the tunnel session is adjustable and there can be a



Shen, et al.             Expires April 27, 2006                [Page 14]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   many-to-one mapping to the tunnel session, related policy mechanism
   needs to decide how the adjustment to the tunnel reservation should
   be done to accommodate the end-to-end sessions mapped onto it.  As
   discussed in [18], there could be multiple choices.  In the first,
   the tunnel reservation is never adjusted, which makes the tunnel a
   rough equivalent of a fixed-capacity hardware link ("hard pipe").  In
   the second, the tunnel reservation is adjusted whenever a new end-to-
   end reservation arrives or an old one is torn down ("soft pipe").
   Doing this will require the Texit to keep track of the resources
   allocated to the tunnel and the resources actually in use by end-to-
   end reservations separately.  It is often appropriate to adopt a
   third choice, where we use some hysteresis in the adjustment of the
   tunnel reservation parameters.  The tunnel reservation is adjusted
   upwards or downwards occasionally, whenever the end-to-end
   reservation level has changed enough to warrant the adjustment.  This
   trades off extra resource usage in the tunnel for reduced control
   traffic and overhead.

6.4.  Protocol Operation for Mixed Signaling Tunnels

   In reality, a tunnel can contain both individual signaling sessions
   and aggregate sessions; a configured tunnel session also does not
   have to be an aggregate session.  Different types of tunnel sessions
   in the mixed tunnel can be dealt with using corresponding mechanisms
   discussed above, the choice of mapping an end-to-end session to a
   specific type of tunnel session is up to policy control.


7.  Message Processing Rules for Selected End-to-End QoS NSLP Messages
    at Tunnel Endpoints

   Following are basic message processing rules for involved end-to-end
   QoS NSLP messages.  More details will be provided in later version of
   this document.

7.1.  End-to-End QUERY Message at Tentry

   When an end-to-end QUERY message is received at Tentry, Tentry checks
   whether the end-to-end session is entitled to tunnel resources.

   If the end-to-end session should be bound to a tunnel session yet to
   be created.  Tentry creates a tunnel QUERY' message and sends it to
   Texit.  Tentry also appends a tunnel BOUND_SESSION_ID object to the
   end-to-end QUERY message.  The tunnel BOUND_SESSION_ID object
   contains the SID of the tunnel session.  The end-to-end QUERY message
   is then encapsulated and sent out through the tunnel interface.

   If the end-to-end session should be bound to an existing tunnel



Shen, et al.             Expires April 27, 2006                [Page 15]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   session (whether aggregate or individual), Tentry appends a tunnel
   BOUND_SESSION_ID object to the end-to-end tunnel QUERY message and
   sends it toward Texit through the tunnel interface.

7.2.  End-to-End QUERY Message at Texit

   When an end-to-end QUERY message containing a tunnel BOUND_SESSION_ID
   object is received, Texit creates a conditional reservation state for
   the end-to-end session.  It also checks to see if a conditional
   reservation state for the associated tunnel session is available.  If
   yes, it reads information from the tunnel session state and sends the
   end-to-end QUERY downstream.  If the conditional reservation state
   for tunnel session is not yet available, it will be created upon
   receiving the tunnel QUERY', and then Texit should forward the end-
   to-end QUERY downstream with information from results of the tunnel
   QUERY'.

   Note that the latest version of QoS NSLP draft [3] defines an R-bit
   in the QUERY message.  More details about dealing with R-bit of the
   QUERY message will be discussed in a later version of this document.

7.3.  End-to-End RESERVE Message at Tentry

   When a RESERVE message is received, in addition to normal processing
   for the request, the following tunnel related functionality is
   performed.

   For sender-initiated RESERVE message,

   If the RESERVE message is received with its T-bit set (RESERVE tear),
   Tentry removes the local state, then encapsulates the RESERVE message
   and tunnels it to Texit.  If there is a tunnel session associated
   with this end-to-end session, Tentry also sends a tunnel RESERVE with
   T-bit set for that tunnel session.

   If the end-to-end RESERVE message is a refresh for an existing end-
   to-end session and this session is associated with a tunnel session,
   the RESERVE message refreshes both two sessions.  If the RESERVE
   message causes changes in resources reserved for the end-to-end
   session, Tentry may create a new tunnel RESERVE' message to change
   the tunnel reservation as well.  At the same time, Tentry appends a
   tunnel BOUND_SESSION_ID object to the end-to-end RESERVE message and
   sends it to Texit through the tunnel interface.

   If the message is the first RESERVE message for an end-to-end
   session, Tentry determines whether the end-to-end session is entitled
   to tunnel resources based on policy control mechanisms outside the
   scope of this document.  If not, no special tunnel related processing



Shen, et al.             Expires April 27, 2006                [Page 16]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   is needed.  Otherwise, if this session should be bound to an existing
   tunnel session (whether aggregate or individual), Tentry creates the
   association between the end-to-end session and the tunnel session.
   Then it appends a tunnel BOUND_SESSION_ID object to the end-to-end
   RESERVE message and sends it through the tunnel interface (i.e. the
   message is encapsulated and tunneled to Texit as normal).

   If the end-to-end session should be bound to a tunnel session yet to
   be created, Tentry assigns the tunnel flow ID, and constructs a
   tunnel RESERVE' message.  The QSPEC in this tunnel RESERVE message
   may be different from the original QSPEC, taking into consideration
   the tunnel overhead of the encapsulation of data packets.  Tentry
   then associates the tunnel session with the end-to-end session in the
   NSLP state and sends the tunnel RESERVE' toward Texit to reserve
   resources over the tunnel.  At the same time, Tentry appends a tunnel
   BOUND_SESSION_ID object to the end-to-end RESERVE message and sends
   it through the tunnel interface.

   For receiver-initiated RESERVE messages,

   If the RESERVE message is received with its T-bit set (RESERVE tear),
   Tentry removes the local state and forwards the message upstream.

   If the end-to-end RESERVE message contains a tunnel BOUND_SESSION_ID
   and is the first end-to-end RESERVE message, Tentry checks whether
   the tunnel session bound to the end-to-end session indicated by the
   RESERVE message already exists.  If yes, Tentry records the
   association between the end-to-end and the tunnel session, reads
   information from the tunnel session to create the end-to-end RESERVE
   message to be forwarded upstream.  If the state for the tunnel
   session is not available yet, Tentry should create state information
   for the tunnel session and indicate that a conditional reservation is
   pending.  When the actual tunnel RESERVE' arrives, the tunnel session
   state will be updated.  If at this time there is a pending
   reservation, Tentry should generate an end-to-end RESERVE message and
   forwards it upstream.

   If the end-to-end RESERVE message contains a tunnel BOUND_SESSION_ID
   and is a refresh, Texit refreshes the end-to-end session.  If the
   RESERVE message causes changes in resources reserved for the end-to-
   end session, Texit checks the state information of the tunnel
   session.  If the reservation has been updated inside the tunnel,
   Texit forwards the RESERVE message toward the sender.  If the tunnel
   reservation update failed, Texit MUST sends a RESPONSE with
   appropriate Error_Spec to the originator of the end-to-end RESERVE
   message.

7.4.  End-to-End RESERVE Message at Texit



Shen, et al.             Expires April 27, 2006                [Page 17]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   When Texit receives a RESERVE message, in addition to normal
   processing of the request, the Texit performs the following steps,

   Sender-initiated RESERVE,

   If the end-to-end RESERVE message is received with its T-bit set
   (RESERVE tear), Texit removes the local state, then forwards the
   RESERVE message downstream.

   If the end-to-end RESERVE message contains a tunnel BOUND_SESSION_ID
   and is the first end-to-end RESERVE message, Texit checks whether the
   state for the tunnel session indicated by the RESERVE message already
   exists.  If yes, Texit records the association between the end-to-end
   and the tunnel session and reads information from the tunnel session
   to create the end-to-end RESERVE message to be forwarded downstream.
   If the state for the tunnel session is not available yet, Texit
   should create state information for the tunnel session and indicate
   that a conditional reservation is pending.  When the actual tunnel
   RESERVE' arrives, the tunnel session state will be updated.  If at
   this time there is a pending reservation, Texit will generate an end-
   to-end RESERVE message and forwards it downstream.

   If the end-to-end RESERVE message contains a tunnel BOUND_SESSION_ID
   and is a refresh, Texit refreshes the end-to-end session.  If the
   RESERVE message causes changes in resources reserved for the end-to-
   end session, Texit checks the state information of the tunnel
   session.  If the reservation has been updated inside the tunnel,
   Texit forwards the RESERVE message toward the receiver.  If the
   tunnel reservation update failed, Texit MUST send a RESPONSE with
   appropriate Error_Spec to the originator of the end-to-end RESERVE
   message.

   Note that the processing rules for end-to-end RESERVE at Texit in
   sender-initiated case is similar to those for end-to-end RESERVE at
   Tentry in receiver-initiated case.

   Receiver-initiated RESERVE,

   If the RESERVE message is received with its T-bit set (RESERVE tear),
   Texit removes the local state, then forwards the RESERVE message
   upstream.  If there is an individual tunnel session associated with
   this end-to-end session, Texit also sends a tunnel RESERVE' with
   T-bit set for that tunnel session.

   Otherwise Texit checks to see if the end-to-end session is associated
   with a tunnel session.  If only conditional reservation state is
   found and no actual reservation has been made, this RESERVE is the
   first end-to-end RESERVE message.  Texit appends a tunnel



Shen, et al.             Expires April 27, 2006                [Page 18]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   BOUND_SESSION_ID object to this end-to-end RESERVE message and sends
   it toward Tentry through the tunnel interface.  Meanwhile it sends
   tunnel RESERVE' message toward Tentry to reserve tunnel resources.

   If the end-to-end session is bound to a tunnel session and the
   RESERVE message is a refresh, it refreshes both the end-to-end
   session and tunnel session.  If the RESERVE message causes changes in
   resources reserved for the end-to-end session, Texit may create a new
   tunnel RESERVE' message to change the tunnel reservation as well.
   Meanwhile, the end-to-end RESERVE is appended with the tunnel
   BOUND_SESSION_ID object and sent to Tentry through the reverse path.

7.5.  Special Processing Rules for Many-to-One Mapping Tunnels

   There are special situations where the end-to-end session is bound to
   pre-configured tunnel sessions in a many-to-one mapping.

   If the associated tunnel session is a "hard pipe" session, arrival of
   a new end-to-end reservation or adjustment of an existing end-to-end
   session may cause the overall resources needed in the tunnel session
   to exceed its capacity, this case is treated as admission control
   failure same as that of a tunnel reservation failure.  Tentry should
   create a RESPONSE message with appropriate Error_Spec and send it to
   the originator of the RESERVE message.

   If the associated tunnel session is a "soft pipe" session, arrival of
   a new end-to-end reservation or adjustment/deletion of existing
   sessions may cause the tunnel session to be modified.  It is
   recommended that some hysteresis is enforced in the adjustment of the
   tunnel reservation parameters.  This requires tunnel endpoint to keep
   track of both the allocated tunnel session resources and the
   resources actually used by end-to-end sessions bound to that tunnel
   session.


8.  Other Considerations

8.1.  Other Types of NSLP

   This document discusses QoS NSLP.  It will be good if the scheme in
   this document could work with other NSLPs as well.  Since NSIS-tunnel
   operation involves specific NSLP itself and different NSLPs have
   different message exchange semantics, the NSIS-tunnel specification
   would not be the same for all NSLPs.  However the basic aspects
   behind NSIS-tunnel operation are indeed similar.  NATFW NSLP is the
   only other main NSLP currently developed by the NSIS working group.
   The most important signaling operation in NATFW NSLP is CREATE.
   Assuming Tentry is a NATFW NSLP, the tunnel-handling for CREATE



Shen, et al.             Expires April 27, 2006                [Page 19]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   operation will be very similar to the sender-initiated QoS
   reservation case.  There are also a number of reverse directional
   operations in NATFW NSLP, e.g., RESERVE_EXTERNAL_ADDRESS, UCREATE.
   It is not very clear whether tunnel will cause problems with these
   messages in general.  But they are likely easier to be dealt with
   than the receiver-initiated reservation case in QoS NSLP.  The next
   version of this document will discuss more on this topic.

8.2.  IPSEC Flows

   If the tunnel supports IPSEC (especially ESP in Tunnel-Mode with or
   without AH), it may use the flow label, TC field, or IPSEC SPI along
   with the tunnel source and destination address, as discussed in
   Section 4.1, to form the tunnel Flow ID.  All these are standard NSIS
   MRI fields that should be matched by the NSIS packet classifier.  We
   may also define virtual destination ports as in [19] to provide
   further flow demultiplexing capability at the destination side if
   necessary.

8.3.  NSIS-Tunnel and Mobility

   The NSIS-tunnel operation needs to interact with IP mobility in an
   efficient way.  For aggregate signaling tunnels, the process is
   relatively straightforward because tunnel session resources are
   usually set up in advance.  For individual signaling tunnels, one way
   to improve tunnel NSIS-mobility efficiency is to keep the SESSION_ID
   of the tunnel session unique when tunnel flow ID changes during
   mobility as illustrated below.

   With a mobile IP tunnel, one tunnel endpoint is the Home Agent (HA),
   and the other is the Mobile Node (MN) if collocated Care-of-Address
   (CoA) is used, or the Foreign Agent (FA) if FA CoA is used.  When MN
   is a receiver, Tentry is the HA and Texit is the MN or FA.  In case
   of a mobility event, handoff tunnel signaling messages will start
   from HA, which may use the same SID for the new tunnel session.  When
   MN is a sender and collocated CoA is used, Tentry is the MN and Texit
   is the HA.  Handoff tunnel signaling is started at the MN.  It may
   also use the SID of the previous tunnel session for the new tunnel
   session.  When MN is a sender and FA CoA is used, the situation is
   complicated, because Tentry has changed from the old FA to the new
   FA.  The new FA does not have the SID of the previous tunnel session.

   When mobile IP is working on a bi-directional tunneling mode, NSIS-
   tunnel operation with mobility may be further improved by localizing
   the handoff tunnel signaling process under the HA (i.e., without
   going through the path between HA and CN).





Shen, et al.             Expires April 27, 2006                [Page 20]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


9.  Security Considerations

   This draft does not draw new security threats.  Security
   considerations for NSIS NTLP and QoS NSLP are discussed in [2] and
   [3] respectively.  General threats for NSIS can be found in [21].


10.  Appendix

10.1.  Summary of RSVP Operation Over IP Tunnels

   RFC 2746 [18] provides an example scheme for RSVP operation over IP
   tunnels.  The scheme needs to be supported by both the Tentry and
   Texit.  To address the tunnel signaling visibility problem, separate
   tunnel signaling sessions are performed for end-to-end sessions.  A
   binding between the tunnel sessions and the end-to-end sessions is
   established.  Both the Tentry and Texit must agree on the binding so
   that changes in the original reservation state can be correctly
   mapped into changes in the tunnel reservation state, and that errors
   reported by intermediate routers to the tunnel endpoints can be
   correctly transformed into errors reported by the tunnel endpoints to
   the end-to-end RSVP session.  To address the tunnel QoS data
   visibility problem, a UDP header is inserted to all QoS data packets
   following the tunnel IP header.  The additional UDP header provides
   source and destination ports that allow intermediate tunnel nodes to
   use standard RSVP filterspec handling and demultiplex different
   tunnel RSVP sessions.

   The RFC 2746 scheme also mentions that in the case where the IP-in-IP
   tunnel supports IPSEC (especially ESP in tunnel-mode with or without
   AH), the tunnel session uses the GPI SESSION and GPI SENDER_TEMPLATE,
   FILTER_SPEC as defined in [19] for the PATH and RESV messages.  Data
   packets are not encapsulated with a UDP header since the SPI can be
   used by the intermediate nodes for classification purposes.

10.2.  Various Design Alternatives

10.2.1.  Carrying Signaling Messages over the Tunnel

   The contents of original end-to-end singling messages are not
   directly examined by tunnel intermediate nodes.  To carry out tunnel
   signaling we choose to generate separate signaling messages for the
   tunnel signaling session.  Another option is to stack tunnel specific
   objects on top of the original end-to-end message and make these
   messages visible to tunnel intermediate nodes so they may serve both
   the end-to-end session and tunnel session.  This turns out to be very
   difficult because the actual tunnel signaling messages differ from
   the end-to-end signaling message both in GIST layer and NSLP layer



Shen, et al.             Expires April 27, 2006                [Page 21]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   information, such as MRI and QSPEC.  Another reason not to choose
   this approach is that tunnel signaling is a process involving message
   exchanges in both directions inside the tunnel.  A separate tunnel
   session signaling is much cleaner.

10.2.2.  Packet Classification over the Tunnel

   Packet classification over the tunnel may be done in either of the
   two ways: first, retaining the end-to-end packet classification
   rules; second, using tunnel specific classification rules.  In the
   first approach, tunnel packet classification is not tied with tunnel
   MRI.  This is a useful characteristic especially in handling tunnel
   mobility - as mobility occurs, the tunnel MRI changes, but the packet
   classification rule does not change.  Therefore, the common path
   after a handover does not need to be updated, resulting in a better
   handoff performance.  The main problem with this approach is that
   most existing routers do not support inspection of inner IP headers
   in an IP tunnel, where the tunnel independent packet classification
   fields usually reside.  Therefore this document chose the second
   approach which does not pose special requirements on intermediate
   tunnel nodes.

10.2.3.  Tunnel Binding Methods

   In this document, the end-to-end session and tunnel session use
   different SESSION_IDs and they are associated with each other using
   the BOUND_SESSION_ID object.  This choice is obvious for aggregate
   signaling tunnels because in that case the original end-to-end
   session and the corresponding aggregate tunnel session require
   independent control.

   Sessions in individual signaling tunnels are created and deleted
   along with the related end-to-end session.  So association between
   the end-to-end session and the corresponding individual tunnel
   session has another choice: the two sessions may share the same
   SESSION_ID, which should be the SESSION_ID of the original end-to-end
   session.  The specific advantage of this choice is that it conforms
   to the general rule that the SESSION_ID should not be modified end-
   to-end [13].  It simplifies the handling of NSIS mobility inside the
   tunnel because the end-to-end session and all associated tunnel
   sessions share the same SESSION_ID.  Problems with this choice arise
   when there is a need to convey the association from one tunnel
   endpoint to the other.  The BOUND_SESSION_ID object cannot be used
   because the SESSION_IDs are the same.  It may be possible to define a
   similar BOUND_FLOW_ID object.  However, since flow ID is usually
   derived from MRI, if a NAT is present in the tunnel, this
   BOUND_FLOW_ID object will have to be modified in the middle, which
   makes the process fairly complicated.  Furthermore, it is not desired



Shen, et al.             Expires April 27, 2006                [Page 22]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   to have different session association mechanisms for aggregate
   signaling tunnels and individual signaling tunnels.  Therefore, we
   decide to use the same BOUND_SESSION_ID mechanism also in individual
   signaling tunnels.  Note that, in this case the mobility handling
   inside the tunnel can still be optimized in certain situations, as
   discussed in Section 8.3

10.2.4.  Tunnel Binding Indication

   In this document we added a Binding_Code field to the existing
   BOUND_SESSION_ID object in order to indicate the nature of binding.
   Two other options considered are:

   1.  Define a designated "tunnel object" to be included when the
       tunnel binding needs to be conveyed.
   2.  Define a "tunnel bit" in corresponding NSLP message headers.

   These options are not chosen because they either need to create
   entirely new object, or need to change basic message headers.  They
   are also not generic solutions that can cover other binding causes.

10.2.5.  Carrying the Tunnel Binding Object

   There are basically two ways to carry the binding object between
   Tentry and Texit, using (a) end-to-end signaling messages or (b)
   tunnel signaling messages.

   Option (a) is cleaner in the sense that only tunnel endpoints are
   involved in this process.  Option (b) embeds the binding information
   in the tunnel signaling messages.  Since both the tunnel SID and flow
   ID are available in the tunnel signaling message, it might even be
   possible to just use a new tunnel bit in the message headers without
   including the binding object.  The disadvantage of option (b) is that
   intermediate tunnel nodes will be exposed to the binding message, and
   modifications to basic message formats may be needed.  Therefore the
   choice in this document is option (a).

10.2.6.  Alternative Ways of End-to-End and Tunnel Session Interaction

   There are many other alternatives of integrating the end-to-end and
   tunnel session signaling.  In general, different approaches can be
   grouped into two modes, sequential mode and parallel mode.  In
   sequential mode, end-to-end signaling pauses when tunnel signaling is
   started, and resumes upon finish of the tunnel signaling; in parallel
   mode, end-to-end signaling continues outside the tunnel while tunnel
   signaling is in process.  We have different ways to define when a
   particular tunnel signaling procedure is completed.  For example, a
   tunnel QUERY' or RESERVE' may be considered finished when the



Shen, et al.             Expires April 27, 2006                [Page 23]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   initiating endpoint receives the corresponding tunnel RESPONSE'
   message.  Compared to the approach we adopted in the main part of
   this document, this approach introduces more signaling delays.

   Tunnel session initiation is another issue that is quite flexible.
   It is also possible to initiate the tunnel session from Texit, or at
   different points of the end-to-end signaling.  For example, the
   tunnel session can be started when Tentry receives the first end-to-
   end RESERVE message, as in the case of [18].  Unlike the scheme we
   presented above, this will not allow the first end-to-end QUERY
   message to trigger a tunnel QUERY' and gather tunnel characteristics
   along with the rest of the end-to-end path.  But the assumption of
   our scheme is that Tentry already knows Texit also supports the NSIS-
   tunnel scheme, so it makes sense to start preparing for tunnel
   session signaling early.

   Tentry always needs to be NSIS-Tunnel aware because it at least needs
   to encapsulate packets into special tunnel flow IDs.  Texit needs to
   be NSIS-tunnel aware if the tunnel reservation is receiver initiated.
   When the tunnel reservation is sender-initiated, it is possible that
   Texit is NSIS-Tunnel unaware and the tunnel signaling still works.
   However, the condition is that no special packet decapsulation is
   needed (e.g. when UDP insertion is used for tunnel flow ID).
   Considering that most of the time we might have a bi-directional
   tunnel and also for more general applicability, we assumed both
   tunnel endpoints to be NSIS-Tunnel aware in this document.


11.  Acknowledgements


12.  References

12.1.  Normative References

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

   [2]  Schulzrinne, H. and R. Hancock, "GIST: General Internet
        Signaling Transport", draft-ietf-nsis-ntlp-08 (work in
        progress), September 2005.

   [3]  Bosch, S., "NSLP for Quality-of-Service signalling",
        draft-ietf-nsis-qos-nslp-08 (work in progress), October 2005.

12.2.  Informative References

   [4]   Hanks, S., Li, T., Farinacci, D., and P. Traina, "Generic



Shen, et al.             Expires April 27, 2006                [Page 24]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


         Routing Encapsulation (GRE)", RFC 1701, October 1994.

   [5]   Hanks, S., Li, T., Farinacci, D., and P. Traina, "Generic
         Routing Encapsulation over IPv4 networks", RFC 1702,
         October 1994.

   [6]   Gilligan, R. and E. Nordmark, "Transition Mechanisms for IPv6
         Hosts and Routers", RFC 2893, August 2000.

   [7]   Perkins, C., "IP Encapsulation within IP", RFC 2003,
         October 1996.

   [8]   Perkins, C., "Minimal Encapsulation within IP", RFC 2004,
         October 1996.

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

   [10]  Kent, S. and R. Atkinson, "IP Encapsulating Security Payload
         (ESP)", RFC 2406, November 1998.

   [11]  Conta, A. and S. Deering, "Generic Packet Tunneling in IPv6
         Specification", RFC 2473, December 1998.

   [12]  Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., and W.
         Weiss, "An Architecture for Differentiated Services", RFC 2475,
         December 1998.

   [13]  Hancock, R., "Next Steps in Signaling: Framework",
         draft-ietf-nsis-fw-07 (work in progress), December 2004.

   [14]  Ash, J., "QoS-NSLP QSPEC Template", draft-ietf-nsis-qspec-06
         (work in progress), October 2005.

   [15]  Stiemerling, M., "NAT/Firewall NSIS Signaling Layer Protocol
         (NSLP)", draft-ietf-nsis-nslp-natfw-07 (work in progress),
         July 2005.

   [16]  Lee, S., "Applicability Statement of NSIS Protocols in Mobile
         Environments",
         draft-ietf-nsis-applicability-mobility-signaling-02 (work in
         progress), July 2005.

   [17]  Farinacci, D., Li, T., Hanks, S., Meyer, D., and P. Traina,
         "Generic Routing Encapsulation (GRE)", RFC 2784, March 2000.

   [18]  Terzis, A., Krawczyk, J., Wroclawski, J., and L. Zhang, "RSVP
         Operation Over IP Tunnels", RFC 2746, January 2000.



Shen, et al.             Expires April 27, 2006                [Page 25]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


   [19]  Berger, L. and T. O'Malley, "RSVP Extensions for IPSEC Data
         Flows", RFC 2207, September 1997.

   [20]  Rajahalme, J., Conta, A., Carpenter, B., and S. Deering, "IPv6
         Flow Label Specification", RFC 3697, March 2004.

   [21]  Tschofenig, H. and D. Kroeselberg, "Security Threats for Next
         Steps in Signaling (NSIS)", RFC 4081, June 2005.











































Shen, et al.             Expires April 27, 2006                [Page 26]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


Authors' Addresses

   Charles Shen
   Columbia University
   Department of Computer Science
   1214 Amsterdam Avenue, MC 0401
   New York, NY  10027
   USA

   Phone: +1 212 854 5599
   Email: charles@cs.columbia.edu


   Henning Schulzrinne
   Columbia University
   Department of Computer Science
   1214 Amsterdam Avenue, MC 0401
   New York, NY  10027
   USA

   Phone: +1 212 939 7004
   Email: schulzrinne@cs.columbia.edu


   Sung-Hyuck Lee
   SAMSUNG Advanced Institute of Technology
   San 14-1, Nongseo-ri, Giheung-eup
   Yongin-si, Gyeonggi-do  449-712
   KOREA

   Phone: +82 31 280 9552
   Email: starsu.lee@samsung.com


   Jong Ho Bang
   SAMSUNG Advanced Institute of Technology
   San 14-1, Nongseo-ri, Giheung-eup
   Yongin-si, Gyeonggi-do  449-712
   KOREA

   Phone: +82 31 280 9585
   Email: jh0278.bang@samsung.com









Shen, et al.             Expires April 27, 2006                [Page 27]

Internet-Draft       NSIS Operation Over IP Tunnels         October 2005


Intellectual Property Statement

   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in BCP 78 and BCP 79.

   Copies of IPR disclosures made to the IETF Secretariat and any
   assurances of licenses to be made available, or the result of an
   attempt made to obtain a general license or permission for the use of
   such proprietary rights by implementers or users of this
   specification can be obtained from the IETF on-line IPR repository at
   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.


Disclaimer of Validity

   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 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.


Copyright Statement

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


Acknowledgment

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




Shen, et al.             Expires April 27, 2006                [Page 28]



PAFTECH AB 2003-20262026-04-22 23:01:33