Tech-invite3GPPspaceIETFspace
96959493929190898887868584838281807978777675747372717069686766656463626160595857565554535251504948474645444342414039383736353433323130292827262524232221201918171615141312111009080706050403020100
in Index   Prev   Next

RFC 5946

Resource Reservation Protocol (RSVP) Extensions for Path-Triggered RSVP Receiver Proxy

Pages: 35
Proposed Standard
Updates:  2205

Top   ToC   RFC5946 - Page 1
Internet Engineering Task Force (IETF)                    F. Le Faucheur
Request for Comments: 5946                                         Cisco
Updates: 2205                                                  J. Manner
Category: Standards Track                               Aalto University
ISSN: 2070-1721                                             A. Narayanan
                                                                   Cisco
                                                              A. Guillou
                                                                     SFR
                                                                H. Malik
                                                                  Airtel
                                                            October 2010


            Resource Reservation Protocol (RSVP) Extensions
                 for Path-Triggered RSVP Receiver Proxy

Abstract

Resource Reservation Protocol (RSVP) signaling can be used to make end-to-end resource reservations in an IP network in order to guarantee the Quality of Service (QoS) required by certain flows. With conventional RSVP, both the data sender and receiver of a given flow take part in RSVP signaling. Yet, there are many use cases where resource reservation is required, but the receiver, the sender, or both, is not RSVP-capable. Where the receiver is not RSVP- capable, an RSVP router may behave as an RSVP Receiver Proxy, thereby performing RSVP signaling on behalf of the receiver. This allows resource reservations to be established on the segment of the end-to- end path from the sender to the RSVP Receiver Proxy. However, as discussed in the companion document "RSVP Proxy Approaches", RSVP extensions are needed to facilitate operations with an RSVP Receiver Proxy whose signaling is triggered by receipt of RSVP Path messages from the sender. This document specifies these extensions. Status of This Memo This is an Internet Standards Track document. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 5741. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc5946.
Top   ToC   RFC5946 - Page 2
Copyright Notice

   Copyright (c) 2010 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

   This document may contain material from IETF Documents or IETF
   Contributions published or made publicly available before November
   10, 2008.  The person(s) controlling the copyright in some of this
   material may not have granted the IETF Trust the right to allow
   modifications of such material outside the IETF Standards Process.
   Without obtaining an adequate license from the person(s) controlling
   the copyright in such materials, this document may not be modified
   outside the IETF Standards Process, and derivative works of it may
   not be created outside the IETF Standards Process, except to format
   it for publication as an RFC or to translate it into languages other
   than English.
Top   ToC   RFC5946 - Page 3

Table of Contents

1. Introduction ....................................................4 1.1. Conventions Used in This Document ..........................7 2. Terminology .....................................................7 3. RSVP Extensions for Sender Notification .........................8 3.1. Sender Notification via PathErr Message ...................11 3.1.1. Composition of SESSION and Sender Descriptor .......14 3.1.2. Composition of ERROR_SPEC ..........................14 3.1.3. Use of Path_State_Removed Flag .....................15 3.1.4. Use of PathErr by Regular Receivers ................16 3.2. Sender Notification via Notify Message ....................17 4. Mechanisms for Maximizing the Reservation Span .................23 4.1. Dynamic Discovery of Downstream RSVP Functionality ........24 4.2. Receiver Proxy Control Policy Element .....................26 4.2.1. Default Handling ...................................29 5. Security Considerations ........................................29 5.1. Security Considerations for the Sender Notification via Notify Message ...........................30 5.2. Security Considerations for the Receiver Proxy Control Policy Element ....................................31 6. IANA Considerations ............................................32 6.1. RSVP Error Codes ..........................................32 6.2. Policy Element ............................................32 7. Acknowledgments ................................................33 8. References .....................................................33 8.1. Normative References ......................................33 8.2. Informative References ....................................34
Top   ToC   RFC5946 - Page 4

1. Introduction

Guaranteed Quality of Service (QoS) for some applications with tight QoS requirements may be achieved by reserving resources in each node on the end-to-end path. The main IETF protocol for these resource reservations is the Resource Reservation Protocol (RSVP), as specified in [RFC2205]. RSVP does not require that all intermediate nodes support RSVP, but it assumes that both the sender and the receiver of the data flow support RSVP. However, there are environments where it would be useful to be able to reserve resources for a flow (at least a subset of the flow path) even when the sender or the receiver (or both) is not RSVP-capable. Since both the data sender and receiver may be unaware of RSVP, there are two types of RSVP Proxies. In the first case, an entity in the network needs to invoke RSVP on behalf of the data sender and thus generate RSVP Path messages, and eventually receive, process, and sink Resv messages. We refer to this entity as the RSVP Sender Proxy. In the second case, an entity in the network needs to operate RSVP on behalf of the receiver and thus receive Path messages sent by a data sender (or by an RSVP Sender Proxy), and reply to those with Resv messages generated on behalf of the data receiver(s). We refer to this entity as the RSVP Receiver Proxy. RSVP Proxy approaches are presented in [RFC5945]. That document also discusses, for each approach, how the reservations controlled by the RSVP Proxy can be synchronized with the application requirements (e.g., when to establish, maintain, and tear down the RSVP reservation to satisfy application requirements). One RSVP Proxy approach is referred to as the Path-Triggered RSVP Receiver Proxy approach. With this approach, the RSVP Receiver Proxy uses the RSVP Path messages generated by the sender (or RSVP Sender Proxy) as the cue for establishing the RSVP reservation on behalf of the non-RSVP-capable receiver(s). The RSVP Receiver Proxy is effectively acting as an intermediary making reservations (on behalf of the receiver) under the sender's control (or RSVP Sender Proxy's control). This somewhat changes the usual RSVP reservation model where reservations are normally controlled by receivers. Such a change greatly facilitates operations in the scenario of interest here, which is where the receiver is not RSVP-capable. Indeed it allows the RSVP Receiver Proxy to remain application-unaware by taking advantage of the application awareness and RSVP awareness of the sender (or RSVP Sender Proxy). Since the synchronization between an RSVP reservation and an application is now effectively performed by the sender (or RSVP Sender Proxy), it is important that the sender (or RSVP Sender Proxy)
Top   ToC   RFC5946 - Page 5
   is aware of the reservation state.  However, as conventional RSVP
   assumes that the reservation is to be controlled by the receiver,
   some notifications about reservation state (notably the error message
   sent in the case of admission control rejection of the reservation)
   are only sent towards the receiver and therefore, in our case, sunk
   by the RSVP Receiver Proxy.  Section 3 of this document specifies
   extensions to RSVP procedures allowing such notifications to be also
   conveyed towards the sender.  This facilitates synchronization by the
   sender (or RSVP Sender Proxy) between the RSVP reservation and the
   application requirements, and it facilitates sender-driven control of
   reservation in scenarios involving a Path-Triggered RSVP Receiver
   Proxy.

   With unicast applications in the presence of RSVP Receiver Proxies,
   if the sender is notified about the state of the reservation towards
   the receiver (as enabled by this document), the sender is generally
   in a good position to synchronize the reservation with the
   application and to perform efficient sender-driven reservation: the
   sender can control the establishment or removal of the reservation
   towards the receiver by sending Path or PathTear messages,
   respectively.  For example, if the sender is notified that the
   reservation for a point-to-point audio session towards the receiver
   is rejected, the sender may trigger rejection of the session at the
   application layer and may issue a PathTear message to remove any
   corresponding RSVP state (e.g., Path states) previously established.

   However, we note that multicast applications do not always coexist
   well with RSVP Receiver Proxies, since sender notification about
   reservation state towards each RSVP Receiver Proxy may not be
   sufficient to achieve tight application-level synchronization by
   multicast senders.  These limitations stem from the fact that
   multicast operation is receiver driven and, while end-to-end RSVP is
   also receiver driven (precisely to deal with multicast efficiently),
   the use of RSVP Receiver Proxies only allows sender-driven
   reservation.  For example, a sender generally is not aware of which
   receivers have joined downstream of a given RSVP Receiver Proxy, or
   even which RSVP Receiver Proxies have joined downstream of a given
   failure point.  Therefore, it may not be possible to support a mode
   of operation whereby a given receiver only joins a group if that
   receiver benefits from a reservation.  Additionally, a sender may
   have no recourse if only a subset of RSVP Receiver Proxies return
   successful reservations (even if application-level signaling runs
   between the sender and receivers), since the sender may not be able
   to correctly identify the set of receivers who do not have
   reservations.  However, it is possible to support a mode of operation
   whereby multicast traffic is transmitted if and only if all receivers
   benefit from a reservation (from sender to their respective RSVP
   Receiver Proxy): the sender can ensure this by sending a PathTear
Top   ToC   RFC5946 - Page 6
   message and stopping transmission whenever it gets a notification for
   reservation reject for one or more RSVP Receiver Proxies.  It is also
   possible to support a mode of operation whereby receivers join
   independently of whether or not they can benefit from a reservation
   (to their respective RSVP Receiver Proxy), but do benefit from a
   reservation whenever the corresponding resources are reservable on
   the relevant path.

   This document discusses extensions to facilitate operations in the
   presence of a Path-Triggered RSVP Receiver Proxy.  As pointed out
   previously, those apply equally whether RSVP signaling is initiated
   by a regular RSVP sender or by an RSVP Sender Proxy (with some means
   to synchronize reservation state with application-level requirements
   that are outside the scope of this document).  For readability, the
   rest of this document discusses operations assuming a regular RSVP
   sender; however, such an operation is equally applicable where an
   RSVP Sender Proxy is used to initiated RSVP signaling on behalf of a
   non-RSVP-capable sender.

   As discussed in [RFC5945], it is important to keep in mind that the
   strongly recommended RSVP deployment model remains end to end as
   assumed in [RFC2205] with RSVP support on the sender and the
   receiver.  The end-to-end model allows the most effective
   synchronization between the reservation and application requirements.
   Also, when compared to the end-to-end RSVP model, the use of RSVP
   Proxies involves additional operational burden and/or imposes some
   topological constraints.  Thus, the purpose of this document is only
   to allow RSVP deployment in special environments where RSVP just
   cannot be used on some senders and/or some receivers for reasons
   specific to the environment.

   Section 4.1.1 of [RFC5945] discusses mechanisms allowing the RSVP
   reservation for a given flow to be dynamically extended downstream of
   an RSVP Proxy whenever possible (i.e., when the receiver is RSVP-
   capable or when there is another RSVP Receiver Proxy downstream).
   This can considerably alleviate the operational burden and the
   topological constraints associated with Path-Triggered RSVP Receiver
   Proxies.  This allows (without corresponding manual configuration) an
   RSVP reservation to dynamically span as much of the corresponding
   flow path as possible, with any arbitrary number of RSVP Receiver
   Proxies on the flow path and whether or not the receiver is RSVP-
   capable.  In turn, this facilitates migration from an RSVP deployment
   model based on Path-Triggered Receiver Proxies to an end-to-end RSVP
   model, since receivers can gradually and independently be upgraded to
   support RSVP and then instantaneously benefit from end-to-end
   reservations.  Section 4 of this document specifies these mechanisms
   and associated RSVP extensions.
Top   ToC   RFC5946 - Page 7

1.1. 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].

2. Terminology

The following terminology is borrowed from [RFC5945] and is used extensively in this document: o RSVP-capable (or RSVP-aware): supporting the RSVP protocol as per [RFC2205]. o RSVP Receiver Proxy: an RSVP-capable router performing, on behalf of a receiver, the RSVP operations that would normally be performed by an RSVP-capable receiver if end-to-end RSVP signaling were used. Note that while RSVP is used upstream of the RSVP Receiver Proxy, RSVP is not used downstream of the RSVP Receiver Proxy. o RSVP Sender Proxy: an RSVP-capable router performing, on behalf of a sender, the RSVP operations that normally would be performed by an RSVP-capable sender if end-to-end RSVP signaling were used. Note that while RSVP is used downstream of the RSVP Sender Proxy, RSVP is not used upstream of the RSVP Sender Proxy. o Regular RSVP Router: an RSVP-capable router that is not behaving as an RSVP Receiver Proxy nor as an RSVP Sender Proxy. Note that the roles of the RSVP Receiver Proxy, RSVP Sender Proxy, and regular RSVP Router are all relative to one unidirectional flow. A given router may act as the RSVP Receiver Proxy for a flow, as the RSVP Sender Proxy for another flow, and as a regular RSVP router for yet another flow. The following terminology is also used in this document: o Regular RSVP sender: an RSVP-capable host behaving as the sender for the considered flow and participating in RSVP signaling in accordance with the sender behavior specified in [RFC2205]. o Regular RSVP receiver: an RSVP-capable host behaving as the receiver for the considered flow and participating in RSVP signaling in accordance with the receiver behavior specified in [RFC2205].
Top   ToC   RFC5946 - Page 8

3. RSVP Extensions for Sender Notification

This section defines extensions to RSVP procedures allowing sender notification of reservation failure. This facilitates synchronization by the sender between RSVP reservation and application requirements in scenarios involving a Path-Triggered RSVP Receiver Proxy. As discussed in [RFC5945], with the Path-Triggered RSVP Receiver Proxy approach, the RSVP router may be configured to use receipt of a regular RSVP Path message as the trigger for RSVP Receiver Proxy behavior. On receipt of the RSVP Path message, the RSVP Receiver Proxy: 1. establishes the RSVP Path state as per regular RSVP processing. 2. identifies the downstream interface towards the receiver. 3. sinks the Path message. 4. behaves as if a corresponding Resv message (on its way upstream from the receiver) was received on the downstream interface. This includes performing admission control on the downstream interface, establishing a Resv state (in the case of successful admission control), and forwarding the Resv message upstream, sending periodic refreshes of the Resv message and tearing down the reservation if the Path state is torn down. Operation of the Path-Triggered Receiver Proxy in the case of a successful reservation is illustrated in Figure 1.
Top   ToC   RFC5946 - Page 9
 |****|        ***        ***        ***        |**********|      |----|
 | S  |--------*r*--------*r*--------*r*--------| RSVP     |------| R  |
 |****|        ***        ***        ***        | Receiver |      |----|
                                                | Proxy    |
                                                |**********|

      --Path---> --Path---> --Path---> --Path--->

      <---Resv-- <---Resv-- <---Resv-- <---Resv--

      ===================RSVP===================>

      ************************************************************>


 |****| RSVP-capable     |----| Non-RSVP-capable        ***
 | S  | Sender           | R  | Receiver                *r* regular RSVP
 |****|                  |----|                         *** router


 ***> media flow

 ==>  segment of flow path benefiting from an RSVP reservation

                     Figure 1: Successful Reservation

   We observe that, in the case of successful reservation, conventional
   RSVP procedures ensure that the sender is notified of the successful
   reservation establishment.  Thus, no extensions are required in the
   presence of a Path-Triggered RSVP Receiver Proxy in the case of
   successful reservation establishment.

   However, in the case of reservation failure, conventional RSVP
   procedures ensure only that the receiver (or the RSVP Receiver Proxy)
   is notified of the reservation failure.  Specifically, in the case of
   an admission control rejection on a regular RSVP router, a ResvErr
   message is sent downstream towards the receiver.  In the presence of
   an RSVP Receiver Proxy, if we simply follow conventional RSVP
   procedures, this means that the RSVP Receiver Proxy is notified of
   the reservation failure, but the sender is not.  Operation of the
   Path-Triggered RSVP Receiver Proxy in the case of an admission
   control failure, assuming conventional RSVP procedures, is
   illustrated in Figure 2.
Top   ToC   RFC5946 - Page 10
 |****|        ***        ***        ***        |**********|      |----|
 | S  |--------*r*--------*r*--------*r*--------| RSVP     |------| R  |
 |****|        ***        ***        ***        | Receiver |      |----|
                                                | Proxy    |
                                                |**********|

      --Path---> --Path---> --Path---> --Path--->

                            <---Resv-- <---Resv--

                            -ResvErr-> -ResvErr->

      ===================RSVP===================>

      ************************************************************>


 |****| RSVP-capable  |----| Non-RSVP-capable   ***
 | S  | Sender        | R  | Receiver           *r* regular RSVP
 |****|               |----|                    *** router

 ***> media flow

 ==>  segment of flow path benefiting from an RSVP reservation

           Figure 2: Reservation Failure with Conventional RSVP

   While the sender could infer reservation failure from the fact that
   it has not received a Resv message after a certain time, there are
   clear benefits to ensuring that the sender gets a prompt, explicit
   notification in the case of reservation failure.  This includes
   faster end-user notification at the application layer (e.g., busy
   signal) and faster application-level reaction (e.g., application-
   level rerouting), as well as faster release of application-level
   resources.

   Section 3.1 defines a method that can be used to achieve sender
   notification of reservation failure.  A router implementation
   claiming compliance with this document MUST support the method
   defined in Section 3.1.

   Section 3.2 defines another method that can be used to achieve sender
   notification of reservation failure.  A router implementation
   claiming compliance with this document MAY support the method defined
   in Section 3.2.
Top   ToC   RFC5946 - Page 11
   In a given network environment, a network administrator may elect to
   use the method defined in Section 3.1, the method defined in
   Section 3.2, or possibly combine the two.

3.1. Sender Notification via PathErr Message

With this method, the RSVP Receiver Proxy MUST generate a PathErr message whenever the two following conditions are met: 1. The reservation establishment has failed (or the previously established reservation has been torn down). 2. The RSVP Receiver Proxy determines that it cannot re-establish the reservation (e.g., by adapting its reservation request in reaction to the error code provided in the received ResvErr in accordance with local policy). Note that this notion of generating a PathErr message upstream in order to notify the sender about a reservation failure is not completely new. It is borrowed from [RFC3209] where it was introduced in order to satisfy a similar requirement, which is to allow an MPLS Traffic Engineering (TE) Label Switching Router to notify the TE Tunnel head-end (i.e., the sender) of a failure to establish (or maintain) a TE Tunnel Label Switch Path. Operation of the Path-Triggered RSVP Receiver Proxy in the case of an admission control failure, using sender notification via a PathErr message, is illustrated in Figure 3.
Top   ToC   RFC5946 - Page 12
 |****|        ***        ***        ***        |**********|      |----|
 | S  |--------*r*--------*r*--------*r*--------| RSVP     |------| R  |
 |****|        ***        ***        ***        | Receiver |      |----|
                                                | Proxy    |
                                                |**********|

      --Path---> --Path---> --Path---> --Path--->

                            <---Resv-- <---Resv--

                            -ResvErr-> -ResvErr->

      <-PathErr- <-PathErr- <-PathErr- <-PathErr-

      ===================RSVP===================>

      ************************************************************>


 |****| RSVP-capable  |----| Non-RSVP-capable   ***
 | S  | Sender        | R  | Receiver           *r* regular RSVP
 |****|               |----|                    *** router

 ***> media flow

 ==>  segment of flow path benefiting from RSVP
      (but not benefiting from a reservation in this case)

    Figure 3: Reservation Failure with Sender Notification via PathErr

   The role of this PathErr is to notify the sender that the reservation
   was not established or was torn down.  This may be in the case of
   receipt of a ResvErr, or because of local failure on the Receiver
   Proxy.  On receipt of a ResvErr, in all situations where the
   reservation cannot be installed, the Receiver Proxy MUST generate a
   PathErr towards the sender.  For local failures on the Receiver Proxy
   node, if a similar failure on an RSVP midpoint would cause the
   generation of a ResvErr (for example, admission control failure), the
   Receiver Proxy MUST generate a PathErr towards the sender.  The
   Receiver Proxy MAY additionally generate a PathErr upon local
   failures that would not ordinarily cause generation of a ResvErr
   message, such as those described in Appendix B of [RFC2205].

   The PathErr generated by the Receiver Proxy corresponds to the
   sender(s) that triggered generation of Resv messages that failed.
   For FF-style (Fixed-Filter) reservations, the Receiver Proxy MUST
   send a PathErr towards the (single) sender matching the failed
   reservation.  For SE-style (Shared-Explicit) reservations, the
Top   ToC   RFC5946 - Page 13
   Receiver Proxy MUST send the PathErr(s) towards the set of senders
   that triggered reservations that failed.  This may be a subset of
   senders sharing the same reservation, in which case the remaining
   senders would have their reservation intact and would not receive a
   PathErr.  In both cases, the rules described in Section 3.1.8 of
   [RFC2205] for generating flow descriptors in ResvErr messages also
   apply when generating sender descriptors in PathErr messages.

   For WF-style (Wildcard-Filter) reservations, it is not always
   possible for the Receiver Proxy to reliably know which sender caused
   the reservation failure.  Therefore, the Receiver Proxy SHOULD send a
   PathErr towards each sender.  This means that all the senders will
   receive a notification that the reservation is not established,
   including senders that did not cause the reservation failure.
   Therefore, the method of sender notification via a PathErr message is
   somewhat overly conservative (i.e., in some cases, rejecting
   reservations from some senders when those could have actually been
   established) when used in combination with the Wildcard-Filter style
   (and when there is more than one sender).

   The sender notification via the PathErr method applies to both
   unicast and multicast sessions.  However, for a multicast session, it
   is possible that reservation failure (e.g., admission control
   failure) in a node close to a sender may cause ResvErr messages to be
   sent to a large group of Receiver Proxies.  These Receiver Proxies
   would, in turn, all send PathErr messages back to the same sender,
   which could cause a scalability issue in some environments.

   From the perspective of the sender, errors that prevent a reservation
   from being set up can be classified in two ways:

   1.  Errors that the sender can attempt to correct.  The error code
       for these errors should explicitly be communicated back to the
       sender.  An example of this is "Code 1: Admission Control
       Failure", because the sender could potentially resend a Path
       message with smaller traffic parameters.

   2.  Errors over which the sender has no control.  For these errors,
       it is sufficient to notify the sender that the reservation was
       not set up successfully.  An example of this is "Code 13: Unknown
       Object", because the sender has no control over the objects
       inserted into the reservation by the Receiver Proxy.

   The PathErr message generated by the Receiver Proxy has the same
   format as regular PathErr messages defined in [RFC2205].  The
   SESSION, ERROR_SPEC, and sender descriptor are composed by the
Top   ToC   RFC5946 - Page 14
   Receiver Proxy as specified in the following subsections.  The
   Receiver Proxy MAY reflect back towards the sender in the PathErr any
   POLICY_DATA objects received in the ResvErr.

3.1.1. Composition of SESSION and Sender Descriptor

The Receiver Proxy MUST insert the SESSION object corresponding to the failed reservation into the PathErr. For FF-style reservations, the Receiver Proxy MUST insert a sender descriptor corresponding to the failed reservation into the PathErr. This is equal to the error flow descriptor in the ResvErr received by the Receiver Proxy. For SE-style reservations, the Receiver Proxy MUST insert a sender descriptor corresponding to the sender triggering the failed reservation into the PathErr. This is equal to the error flow descriptor in the ResvErr received by the Receiver Proxy. If multiple flow descriptors could not be admitted at a midpoint node, that node would generate multiple ResvErr messages towards the receiver as per Section 3.1.8 of [RFC2205]. Each ResvErr would contain an error flow descriptor that matches a specific sender. The Receiver Proxy MUST generate a PathErr for each ResvErr received towards the corresponding sender. As specified earlier, for WF-style reservations, the Receiver Proxy SHOULD send a PathErr to each sender.

3.1.2. Composition of ERROR_SPEC

The Receiver Proxy MUST compose the ERROR_SPEC to be inserted into the PathErr as follows: 1. If the Receiver Proxy receives a ResvErr with either of these error codes -- "Code 1: Admission Control Failure" or "Code 2: Policy Control Failure" -- then the Receiver Proxy copies the error code and value from the ERROR_SPEC in the ResvErr into the ERROR_SPEC of the PathErr message. The error node in the PathErr MUST be set to the address of the Receiver Proxy. This procedure MUST also be followed for a local error on the Receiver Proxy that would ordinarily cause a midpoint to generate a ResvErr with one of the above codes. 2. If the Receiver Proxy receives a ResvErr with any error code except the ones listed in item 1 above, it composes a new ERROR_SPEC with error code "36: Unrecoverable Receiver Proxy Error". The error node address in the PathErr MUST be set to the address of the Receiver Proxy. This procedure MUST also be followed for a local error on the Receiver Proxy that would ordinarily cause a midpoint to generate a ResvErr with any error code other than those listed in item 1 above, or if the Receiver Proxy generates a PathErr for a local error that ordinarily would
Top   ToC   RFC5946 - Page 15
       not cause generation of a ResvErr.  In some cases, it may be
       predetermined that the PathErr will not reach the sender.  For
       example, a node receiving a ResvErr with "Code 3: No Path for
       Resv", knows a priori that the PathErr message it generates
       cannot be forwarded by the same node that could not process the
       Resv.  Nevertheless, the procedures above MUST be followed.  For
       the error code "36: Unrecoverable Receiver Proxy Error", the 16
       bits of the Error Value field are:

       *  hhhh hhhh llll llll

       where the bits are:

       *  hhhh hhhh = 0000 0000: then the low order 8 bits (llll llll)
          MUST be set by Receiver Proxy to 0000 0000 and MUST be ignored
          by the sender.

       *  hhhh hhhh = 0000 0001: then the low order 8 bits (llll llll)
          MUST be set by the Receiver Proxy to the value of the error
          code received in the ResvErr ERROR_SPEC (or, in case the
          Receiver Proxy generated the PathErr without having received a
          ResvErr, to the error code value that would have been included
          by the Receiver Proxy in the ERROR_SPEC in similar conditions
          if it was to generate a ResvErr).  This error value MAY be
          used by the sender to further interpret the reason for the
          reservation failure.

       *  hhhh hhhh = any other value: reserved.

   3.  If the Receiver Proxy receives a ResvErr with the InPlace flag
       set in the ERROR_SPEC, it MUST also set the InPlace flag in the
       ERROR_SPEC of the PathErr.

3.1.3. Use of Path_State_Removed Flag

[RFC3473] defines an optional behavior whereby a node forwarding a PathErr message can remove the Path state associated with the PathErr message and indicate so by including the Path_State_Removed flag in the ERROR_SPEC object of the PathErr message. This can be used in some situations to expedite release of resources and minimize signaling load. This section discusses aspects of the use of the Path_State_Removed flag that are specific to the RSVP Receiver Proxy. In any other aspects, the Path_State_Removed flag operates as per [RFC3473].
Top   ToC   RFC5946 - Page 16
   By default, the RSVP Receiver Proxy MUST NOT include the
   Path_State_Removed flag in the ERROR_SPEC of the PathErr message.
   This ensures predictable operations in all environments including
   those where some RSVP routers do not understand the
   Path_State_Removed flag.

   The RSVP Receiver Proxy MAY support an OPTIONAL mode (to be activated
   by configuration) whereby the RSVP Receiver Proxy includes the
   Path_State_Removed flag in the ERROR_SPEC of the PathErr message and
   removes its local Path state.  When all routers on the path of a
   reservation support the Path_State_Removed flag, its use will indeed
   result in expedited resource release and reduced signaling.  However,
   if there are one or more RSVP routers on the path of the reservation
   that do not support the Path_State_Removed flag (we refer to such
   routers as "old RSVP routers"), the use of the Path_State_Removed
   flag will actually result in slower resource release and increased
   signaling.  This is because the Path_State_Removed flag will be
   propagated upstream by an old RSVP router (even if it does not
   understand it and does not tear its Path state).  Thus, the sender
   will not send a Path Tear, and the old RSVP router will release its
   Path state only through refresh time-out.  A network administrator
   needs to keep these considerations in mind when deciding whether to
   activate the use of the Path_State_Removed flag on the RSVP Receiver
   Proxy.  In a controlled environment where all routers are known to
   support the Path_State_Removed flag, its use can be safely activated
   on the RSVP Receiver Proxy.  In other environments, the network
   administrator needs to assess whether the improvement achieved with
   some reservations outweighs the degradation experienced by other
   reservations.

3.1.4. Use of PathErr by Regular Receivers

Note that while this document specifies that an RSVP Receiver Proxy generates a PathErr upstream in the case of reservation failure, this document does NOT propose that the same be done by regular receivers. In other words, this document does NOT propose modifying the behavior of regular receivers as currently specified in [RFC2205]. The rationale for this includes the following: o When the receiver is RSVP-capable, the current receiver-driven model of [RFC2205] is fully applicable because the receiver can synchronize RSVP reservation state and application state (since it participates in both). The sender(s) need not be aware of the RSVP reservation state. Thus, we can retain the benefits of receiver-driven operations that were explicitly sought by [RFC2205], which states, "In order to efficiently accommodate large groups, dynamic group membership, and heterogeneous receiver requirements, RSVP makes receivers responsible for requesting a
Top   ToC   RFC5946 - Page 17
      specific QoS".  But even for the simplest single_sender/
      single_receiver reservations, the current receiver-driven model
      reduces signaling load and per-hop RSVP processing by not sending
      any error message to the sender in case of admission control
      reject.

   o  The motivation for adding sender error notification in the case of
      an RSVP Receiver Proxy lies in the fact that the actual receiver
      can no longer synchronize the RSVP reservation with application
      state (since the receiver does not participate in RSVP signaling),
      while the sender can.  This motivation does not apply in the case
      of a regular receiver.

   o  There is a lot of existing code and deployed systems successfully
      working under the current [RFC2205] model in the absence of proxy
      today.  The current behavior of existing deployed systems should
      not be changed unless there were a very strong motivation.

3.2. Sender Notification via Notify Message

The OPTIONAL method for sender notification of reservation failure defined in this section aims to provide a more efficient method than the one defined in Section 3.1. Its objectives include: o allowing the failure notification to be sent directly upstream to the sender by the router where the failure occurs (as opposed to first traveling downstream towards the Receiver Proxy and then traveling upstream from the Receiver Proxy to the sender, as effectively happens with the method defined in Section 3.1). o allowing the failure notification to travel without hop-by-hop RSVP processing. o ensuring that such a notification is sent to senders that are capable and willing to process it (i.e., to synchronize reservation status with application). o ensuring that such a notification is only sent in case the receiver is not itself capable and willing to do the synchronization with the application (i.e., because we are in the presence of a Receiver Proxy so that RSVP signaling is not visible to the receiver). Note, however, that such benefits come at the cost of: o a requirement for RSVP routers and senders to support the Notify messages and procedures defined in [RFC3473].
Top   ToC   RFC5946 - Page 18
   o  a requirement for senders to process Notify messages traveling
      upstream but conveying a downstream notification.

   [RFC3473] defines (in Section 4.3, "Notify Messages") the Notify
   message that provides a mechanism to inform non-adjacent nodes of
   events related to the RSVP reservation.  The Notify message differs
   from the error messages defined in [RFC2205] (i.e., PathErr and
   ResvErr messages) in that it can be "targeted" to a node other than
   the immediate upstream or downstream neighbor and that it is a
   generalized notification mechanism.  Notify messages are normally
   generated only after a Notify Request object has been received.

   This section discusses aspects of the use of the Notify message that
   are specific to the RSVP Receiver Proxy.  In any other aspects, the
   Notify message operates as per [RFC3473].

   In order to achieve sender notification of reservation failure in the
   context of this document:

   o  An RSVP sender interested in being notified of reservation failure
      MUST include a Notify Request object (containing the sender's IP
      address) in the Path messages it generates.

   o  Upon receiving a Path message with a Notify Request object, the
      RSVP Receiver Proxy MUST include a Notify Request object in the
      Resv messages it generates.  This Notify Request object MUST
      contain either:

      *  the address that was included in the Notify Request of the
         received Path message, a.k.a. the sender's address.  We refer
         to this approach as the "Direct Notify" approach, or

      *  an address of the Receiver Proxy.  We refer to this approach as
         the "Indirect Notify" approach.

   o  Upon receiving a downstream error notification (whether in the
      form of a Notify, ResvErr, or both), the RSVP Receiver Proxy:

      *  MUST generate a Notify message with upstream notification to
         the corresponding sender, if the sender included a Notify
         Request object in its Path messages and if Indirect
         Notification is used.

      *  SHOULD generate a Notify message with upstream notification to
         the corresponding sender, if the sender included a Notify
         Request object in its Path messages and if Direct Notification
         is used.  The reason for this recommendation is that the
         failure node may not support Notify, so that even if Direct
Top   ToC   RFC5946 - Page 19
         Notification was requested by the RSVP Receiver Proxy, the
         sender may not actually have received a Notify from the failure
         node: generating a Notify from the Receiver Proxy will
         accelerate sender notification, as compared to simply relying
         on PathErr, in this situation.  In controlled environments
         where all the nodes are known to support Notify, the Receiver
         Proxy MAY be configured to not generate the Notify with
         upstream notification when Direct Notification is used, in
         order to avoid duplication of Notify messages (i.e., the sender
         receiving both a Notify from the failure node and from the
         Receiver Proxy).

   As a result of these sender and Receiver Proxy behaviors, as per
   existing Notify procedures, if an RSVP router detects an error
   relating to a Resv state (e.g., admission control rejection after IP
   reroute), the RSVP router will send a Notify message (conveying the
   downstream notification with the ResvErr error code) to the IP
   address contained in the Resv Notify Request object.  If this address
   has been set by the RSVP Receiver Proxy to the sender's address
   (Direct Notify), the Notify message is sent directly to the sender.
   If this address has been set by the RSVP Receiver Proxy to one of its
   own addresses (Indirect Notify), the Notify message is sent to the
   RSVP Receiver Proxy that, in turn, will generate a Notify message
   directly addressed to the sender.

   Operation of the Path-Triggered RSVP Receiver Proxy in the case of an
   admission control failure, using sender notification via Direct
   Notify, is illustrated in Figure 4.
Top   ToC   RFC5946 - Page 20
 |****|        ***        ***        ***        |**********|      |----|
 | S  |--------*r*--------*r*--------*r*--------| RSVP     |------| R  |
 |****|        ***        ***        ***        | Receiver |      |----|
                                                | Proxy    |
                                                |**********|

      --Path*--> --Path*--> --Path*--> --Path*-->

                            <--Resv*-- <--Resv*--

      <------NotifyD--------

                            -ResvErr-> -ResvErr->

      <------------------NotifyU------------------

      <-PathErr- <-PathErr- <-PathErr- <-PathErr-

      ===================RSVP===================>

      ************************************************************>


 |****| RSVP-capable  |----| Non-RSVP-capable   ***
 | S  | Sender        | R  | Receiver           *r* regular RSVP
 |****|               |----|                    *** router

 ***> media flow

 ==>  segment of flow path benefiting from RSVP
      (but not benefiting from a reservation in this case)

 Path*  = Path message containing a Notify Request object
          with sender IP Address

 Resv*  = Resv message containing a Notify Request object
          with sender IP address

 NotifyD = Notify message containing a downstream notification

 NotifyU = Notify message containing an upstream notification

          Figure 4: Reservation Failure with Sender Notification
                             via Direct Notify

   Operation of the Path-Triggered RSVP Receiver Proxy in the case of an
   admission control failure, using sender notification via Indirect
   Notify, is illustrated in Figure 5.
Top   ToC   RFC5946 - Page 21
 |****|        ***        ***        ***        |**********|      |----|
 | S  |--------*r*--------*r*--------*r*--------| RSVP     |------| R  |
 |****|        ***        ***        ***        | Receiver |      |----|
                                                | Proxy    |
                                                |**********|

      --Path*--> --Path*--> --Path*--> --Path*-->

                            <--Resv*-- <--Resv*--

                            -------NotifyD------->

      <------------------NotifyU------------------

                            -ResvErr-> -ResvErr->

      <-PathErr- <-PathErr- <-PathErr- <-PathErr-

      ===================RSVP===================>

      ************************************************************>


 |****| RSVP-capable  |----| Non-RSVP-capable   ***
 | S  | Sender        | R  | Receiver           *r* regular RSVP
 |****|               |----|                    *** router

 ***> media flow

 ==>  segment of flow path benefiting from RSVP
      (but not benefiting from a reservation in this case)

 Path*  = Path message containing a Notify Request object
          with sender IP Address

 Resv*  = Resv message containing a Notify Request object
          with RSVP Receiver Proxy IP address

 NotifyD = Notify message containing a downstream notification

 NotifyU = Notify message containing an upstream notification

          Figure 5: Reservation Failure with Sender Notification
                            via Indirect Notify

   For local failures on the Receiver Proxy node, if a similar failure
   on an RSVP midpoint would cause the generation of a ResvErr (for
   example, admission control failure), the Receiver Proxy MUST generate
Top   ToC   RFC5946 - Page 22
   a Notify towards the sender.  The Receiver Proxy MAY additionally
   generate a Notify upon local failures that would not ordinarily cause
   generation of a ResvErr message, such as those described in
   Appendix B of [RFC2205].

   When the method of sender notification via a Notify message is used,
   it is RECOMMENDED that the RSVP Receiver Proxy also issue a sender
   notification via a PathErr message.  This maximizes the chances that
   the notification will reach the sender in all situations (e.g., even
   if some RSVP routers do not support the Notify procedure, or if a
   Notify message gets dropped).  However, for controlled environments
   (e.g., where all RSVP routers are known to support Notify procedures)
   and where it is desirable to minimize the volume of signaling, the
   RSVP Receiver Proxy MAY rely exclusively on sender notification via a
   Notify message and thus not issue sender notification via a PathErr
   message.

   In environments where there are both RSVP-capable receivers and RSVP
   Receiver Proxies acting on behalf of non-RSVP-capable receivers, a
   sender does not know whether a given reservation is established with
   an RSVP-capable receiver or with an RSVP Receiver Proxy.  Thus, a
   sender that supports the procedures defined in this section may
   include a Notify Request object in Path messages for a reservation
   that happens to be controlled by an RSVP-capable receiver.  This
   document does not define, nor expect, any change in existing receiver
   behavior.  As a result, in this case, the sender will not receive
   Notify messages conveying downstream notifications.  However, this is
   perfectly fine because the synchronization between the RSVP
   reservation state and the application requirement can be performed by
   the actual receiver in this case as per the regular end-to-end RSVP
   model, so that in this case, the sender need not care about
   downstream notifications.

   A sender that does not support the procedures defined in this section
   might include a Notify Request object in Path messages for a
   reservation simply because it is interested in getting upstream
   notifications faster.  If the reservation is controlled by an RSVP
   Receiver Proxy supporting the procedures defined in this section, the
   sender will also receive unexpected Notify messages containing
   downstream notifications.  It is expected that such a sender will
   simply naturally drop such downstream notifications as invalid.
   Because it is RECOMMENDED above that the RSVP Receiver Proxy also
   issue a sender notification via a PathErr message even when sender
   notification is effected via a Notify message, the sender will still
   be notified of a reservation failure in accordance with the "sender
   notification via PathErr" method.  In summary, activating the
   OPTIONAL "sender notification via Notify" method on a Receiver Proxy
   does not prevent a sender that does not support this method from
Top   ToC   RFC5946 - Page 23
   relying on the MANDATORY "sender notification via PathErr" method.
   It would, however, allow a sender supporting the "sender notification
   via Notify" method to take advantage of this OPTIONAL method.

   With Direct Notification, the downstream notification generated by
   the RSVP router where the failure occurs is sent to the IP address
   contained in the Notification Request Object of the corresponding
   Resv message.  In the presence of multiple senders towards the same
   session, it cannot be generally assumed that a separate Resv message
   is used for each sender (in fact, with WF and SE there is a single
   Resv message for all senders, and with FF the downstream router has
   the choice of generating separate Resv messages or a single one).
   Hence, in the presence of multiple senders, Direct Notification
   cannot guarantee notification of all affected senders.  Therefore,
   Direct Notification is better suited to single-sender applications.

   With Indirect Notification, the RSVP Receiver Proxy can generate
   Notify messages with the same logic that is used to generate PathErr
   messages in the "Sender Notification via PathErr" method (in fact,
   those are conveying the same error information, only the Notify is
   directly addressed to the sender while the PathErr travels hop-by-
   hop).  Therefore, operations of the Indirect Notify method in the
   presence of multiple senders is similar to that of the PathErr method
   as discussed in Section 3.1: with FF or SE, a Notify MUST be sent to
   the sender or the set of affected senders, respectively.  With WF,
   the RSVP Receiver Proxy SHOULD send a Notify to each sender, again
   resulting in a somewhat overly conservative behavior in the presence
   of multiple senders.

4. Mechanisms for Maximizing the Reservation Span

This section defines extensions to RSVP procedures allowing an RSVP reservation to span as much of the flow path as possible, with any arbitrary number of RSVP Receiver Proxies on the flow path and whether or not the receiver is RSVP-capable. This facilitates deployment and operations of Path-Triggered RSVP Receiver Proxies since it alleviates the topological constraints and/or configuration load otherwise associated with Receiver Proxies (e.g., make sure there is no RSVP Receiver Proxy for a flow upstream of a given Receiver Proxy, ensure there is no Receiver Proxy for a flow if the receiver is RSVP-capable). This also facilitates migration from an RSVP deployment model based on Path-Triggered Receiver Proxies to an end-to-end RSVP model, since receivers can gradually and independently be upgraded to support RSVP and then instantaneously benefit from end-to-end reservations.
Top   ToC   RFC5946 - Page 24
   Section 4.1 defines a method that allows a Path-Triggered Receiver
   Proxy function to discover whether there is another Receiver Proxy or
   an RSVP-capable receiver downstream and then dynamically extend the
   span of the RSVP reservation downstream.  A router implementation
   claiming compliance with this document SHOULD support the method
   defined in Section 4.1.

   Section 4.2 defines a method that allows a sender to control whether
   or not an RSVP router supporting the Path-Triggered Receiver Proxy
   function is to behave as a Receiver Proxy for a given flow.  A router
   implementation claiming compliance with this document MAY support the
   method defined in Section 4.2.

   In a given network environment, a network administrator may elect to
   use the method defined in Section 4.1, or the method defined in
   Section 4.2, or possibly combine the two.

4.1. Dynamic Discovery of Downstream RSVP Functionality

When generating a proxy Resv message upstream, a Receiver Proxy supporting dynamic discovery of downstream RSVP functionality MUST forward the Path message downstream instead of terminating it (unless dynamic discovery of downstream RSVP functionality is explicitly disabled). If the destination endpoint supports RSVP (or there is another Receiver Proxy downstream), it will receive the Path and generate a Resv upstream. When this Resv message reaches the Receiver Proxy, it recognizes the presence of an RSVP-capable receiver (or of another RSVP Receiver Proxy) downstream and MUST internally convert its state from a proxied reservation to a regular midpoint RSVP behavior. From then on, the RSVP router MUST behave as a regular RSVP router for that reservation (i.e., as if the RSVP router never behaved as an RSVP Receiver Proxy for that flow). This method is illustrated in Figure 6.
Top   ToC   RFC5946 - Page 25
      |****|         ***         |**********|   |----|
      | S  |---------*r*---------| RSVP     |---| R1 |
      |****|         ***         | Receiver |   |----|
                                 | Proxy    |
                                 |          |
                                 |          |            |****|
                                 |          |------------| R2 |
                                 |**********|            |****|

           ---Path--->  --Path--->
              (R1)        (R1)    \-------Path-->
                                  /       (R1)
           <--Resv---  <---Resv---

          ================RSVP===>

          **************************************>


           ---Path--->  --Path--->
              (R2)        (R2)    \-------------Path---->
                                  /             (R2)
           <--Resv---  <---Resv---
                                             <----Resv---

          ================RSVP===========================>

          ***********************************************>


   |****| RSVP-capable  |----| non-RSVP-capable  |****| RSVP-capable
   | S  | Sender        | R  | Receiver          | R  | Receiver
   |****|               |----|                   |****|

   ***
   *r* regular RSVP
   *** router

   (R1) = Path message contains a Session object whose destination is R1

   ***> media flow

   ==>  segment of flow path protected by RSVP reservation

       Figure 6: Dynamic Discovery of Downstream RSVP Functionality
Top   ToC   RFC5946 - Page 26
   If there is no RSVP-capable receiver (or other Receiver Proxy)
   downstream of the Receiver Proxy, then the Path messages sent by the
   Receiver Proxy every RSVP refresh interval (e.g., 30 seconds by
   default) will never be responded to.  However, these messages consume
   a small amount of bandwidth, and in addition would install some RSVP
   state on RSVP-capable midpoint nodes downstream of the first Receiver
   Proxy.  This is seen as a very minor sub-optimality; however, to
   mitigate this, the Receiver Proxy MAY tear down any unanswered
   downstream Path state after a predetermined time (that SHOULD be
   greater or equal to the Path refresh interval), and MAY stop sending
   Path messages for the flow (or MAY only send them at much lower
   frequency).

   This approach only requires support of the behavior described in the
   previous paragraph and does not require any new RSVP extensions.

4.2. Receiver Proxy Control Policy Element

[RFC2750] defines extensions for supporting generic policy-based admission control in RSVP. These extensions include the standard format of POLICY_DATA objects and a description of RSVP handling of policy events. The POLICY_DATA object contains one or more policy elements, each representing a different (and perhaps orthogonal) policy. As an example, [RFC3181] specifies the preemption priority policy element. This document defines a new policy element called the Receiver Proxy Control policy element. This document only defines the use of this policy element in Path messages and for unicast reservations. Other usage is outside the scope of this document. The format of the Receiver Proxy Control policy element is as shown in Figure 7: 0 0 0 1 1 2 2 3 0 . . . 7 8 . . . 5 6 . . . 3 4 . . . 1 +-------------+-------------+-------------+-------------+ | Length | P-Type=REC_PROXY_CONTROL | +-------------+-------------+-------------+-------------+ | Reserved |Control-Value| +---------------------------+---------------------------+ Figure 7: Receiver Proxy Control Policy Element
Top   ToC   RFC5946 - Page 27
   where:

   o  Length: 16 bits

      *  Always 8.  The overall length of the policy element, in bytes.

   o  P-Type: 16 bits

      *  REC_PROXY_CONTROL = 0x07 (see the "IANA Considerations"
         section).

   o  Reserved: 24 bits

      *  SHALL be set to zero on transmit and SHALL be ignored on
         reception.

   o  Control-Value: 8 bits (unsigned)

      *  0 (Reserved): An RSVP Receiver Proxy that understands this
         policy element MUST ignore the policy element if its Control-
         Value is set to that value.

      *  1 (Receiver-Proxy-Needed): An RSVP Receiver Proxy that
         understands this policy element MUST attempt to insert itself
         as a Receiver Proxy for that flow if the corresponding Path
         message contains this Control-Value.  If the Receiver Proxy
         also supports dynamic discovery of downstream RSVP
         functionality as specified in Section 4.1, it MUST still send
         the Path message downstream and attempt to extend the
         reservation downstream so that the reservation can be extended
         to the last Receiver Proxy).  An RSVP sender MAY insert the
         Receiver Proxy Control policy element with this Control-Value
         when it knows (say, by other means, such as application-level
         signaling) that the receiver is not RSVP-capable.

      *  2 (Receiver-Proxy-Not-Needed): An RSVP Receiver Proxy that
         understands this policy element MUST NOT attempt to insert
         itself as a Receiver Proxy for that flow if the corresponding
         Path message contains this Control-Value.  An RSVP sender MAY
         insert the Receiver Proxy Control policy element with this
         Control-Value when it knows (say, by other means, such as
         application-level signaling) that the receiver is RSVP-capable.

   Figure 8 illustrates the method based on the Receiver Proxy Control
   policy element that allows a sender to control whether or not an RSVP
   router supporting the Path-Triggered Receiver Proxy function is to
   behave as a Receiver Proxy for a given flow.
Top   ToC   RFC5946 - Page 28
      |****|         ***         |**********|   |----|
      | S  |---------*r*---------| RSVP     |---| R1 |
      |****|         ***         | Receiver |   |----|
                                 | Proxy    |
                                 |          |
                                 |          |            |****|
                                 |          |------------| R2 |
                                 |**********|            |****|

           ---Path--->  --Path--->
            (R1/N)      (R1/N)

           <--Resv---  <---Resv---

          ================RSVP===>

          **************************************>


           ---Path--->  --Path--->          ----Path---->
            (R2/NN)      (R2/NN)               (R2/NN)

           <--Resv---  <---Resv---          <----Resv----

          ================RSVP===========================>

          ***********************************************>


   |****| RSVP-capable  |----| non-RSVP-capable  |****| RSVP-capable
   | S  | Sender        | R  | Receiver          | R  | Receiver
   |****|               |----|                   |****|

   ***
   *r* regular RSVP
   *** router
   (R1) = Path message contains a Session object whose destination is R1

   (N)  = Path message contains a Receiver Proxy Control policy element
        whose Control-Value is set to Receiver-Proxy-Needed

   (NN) = Path message contains a Receiver Proxy Control policy element
        whose Control-Value is set to Receiver-Proxy-Not-Needed

   ***> media flow
   ==>  segment of flow path protected by RSVP reservation

                Figure 8: Receiver Proxy Control by Sender
Top   ToC   RFC5946 - Page 29

4.2.1. Default Handling

As specified in Section 4.2 of [RFC2750], Policy-Ignorant Nodes (PINs) implement a default handling of POLICY_DATA objects ensuring that those objects can traverse PINs in transit from one Policy Enforcement Point (PEP) to another. This applies to the situations in which POLICY_DATA objects contain the Receiver Proxy Control policy element specified in this document, so that those objects can traverse PINs. Section 4.2 of [RFC2750] also defines a similar default behavior for policy-capable nodes that do not recognize a particular policy element. This applies to the Receiver Proxy Control policy element specified in this document, so that messages carrying the element can traverse policy-capable nodes that do not support the extensions described in this document.

5. Security Considerations

As this document defines extensions to RSVP, the security considerations of RSVP apply, which are discussed in [RFC2205], [RFC4230], and [SEC-GRP-KEY]. Approaches for addressing those concerns are discussed further below. The RSVP authentication mechanisms defined in [RFC2747] and [RFC3097] protect RSVP message integrity hop-by-hop and provide node authentication as well as replay protection, thereby protecting against corruption and spoofing of RSVP messages. These hop-by-hop integrity mechanisms can be used to protect RSVP reservations established using an RSVP Receiver Proxy in accordance with this document. [SEC-GRP-KEY] discusses key types and key provisioning methods as well as their respective applicability to RSVP authentication. RSVP authentication (defined in [RFC2747] and [RFC3097]) SHOULD be supported by an implementation of this document. [SEC-GRP-KEY] also discusses applicability of IPsec mechanisms ([RFC4302], [RFC4303]) and associated key provisioning methods for security protection of RSVP. This discussion applies to the protection of RSVP in the presence of Path-Triggered RSVP Receiver Proxies as defined in this document. A subset of RSVP messages are signaled with the IP router alert option ([RFC2113], [RFC2711]). Based on the current security concerns associated with the use of the IP router alert option, the applicability of RSVP (and therefore of the RSVP Proxy approaches discussed in this document) is limited to controlled environments
Top   ToC   RFC5946 - Page 30
   (i.e., where the security risks associated with the use of the IP
   router alert option are understood and protected against).  The
   security aspects and common practices around the use of the current
   IP router alert option, and consequences of using the IP router alert
   option by applications such as RSVP, are discussed in detail in
   [RTR-ALERT].

   When an RSVP Receiver Proxy is used, the RSVP reservation is no
   longer controlled by the receiver, but rather is controlled by the
   Receiver Proxy (using hints received from the sender in the Path
   message) on behalf of the sender.  Thus, the Receiver Proxy ought to
   be trusted by the end-systems to control the RSVP reservation
   appropriately.  However, basic RSVP operation already assumes a trust
   model where end-systems trust RSVP nodes to appropriately perform
   RSVP reservations.  So the use of an RSVP Receiver Proxy is not seen
   as introducing any significant additional security threat or as
   modifying the RSVP trust model.

   In fact, there are situations in which the use of an RSVP Receiver
   Proxy reduces the security risks.  One example is where a network
   operator relies on RSVP to perform resource reservation and admission
   control within a network and where RSVP senders and RSVP routers are
   located in the operator's premises, while the many RSVP receivers are
   located in the operator's customers' premises.  Such an environment
   is further illustrated in Appendix A.1, "RSVP-Based VoD Admission
   Control in Broadband Aggregation Networks", of [RFC5945].  From the
   operator's perspective, the RSVP routers and RSVP senders are in
   physically secured locations and therefore exposed to a lower risk of
   being tampered with, while the receivers are in locations that are
   physically unsecured and therefore subject to a higher risk of being
   tampered with.  The use of an RSVP Receiver Proxy function
   effectively increases the security of the operator's reservation and
   admission control solution by completely excluding receivers from its
   operation.  Filters can be placed at the edge of the operator
   network, discarding any RSVP message received from end-users.  This
   provides a very simple and effective protection of the RSVP
   reservation and admission control solution operating inside the
   operator's network.

5.1. Security Considerations for the Sender Notification via Notify Message

This document defines, in Section 3.2, an optional method relying on the use of the Notify message specified in [RFC3473]. The Notify message can be sent in a non-hop-by-hop fashion that precludes the use of the RSVP hop-by-hop integrity and authentication model. The approaches and considerations for addressing this issue presented in the Security Considerations section of [RFC3473] apply. In
Top   ToC   RFC5946 - Page 31
   particular, where the Notify messages are transmitted non-hop-by-hop
   and the same level of security provided by [RFC2747] is desired,
   IPsec-based integrity and authentication can be used ([RFC4302] or
   [RFC4303]).  Alternatively, the sending of non-hop-by-hop Notify
   messages can be disabled.  Finally, [SEC-GRP-KEY] discusses the
   applicability of group keying for non-hop-by-hop Notify messages.

5.2. Security Considerations for the Receiver Proxy Control Policy Element

This document also defines, in Section 4.2, the optional Receiver Proxy Control policy element. Policy elements are signaled by RSVP through encapsulation in a Policy Data object as defined in [RFC2750]. Therefore, like any other policy elements, the integrity of the Receiver Proxy Control policy element can be protected as discussed in Section 6 of [RFC2750] by two optional security mechanisms. The first mechanism relies on the RSVP authentication discussed above that provides a chain of trust when all RSVP nodes are policy capable. With this mechanism, the INTEGRITY object is carried inside RSVP messages. The second mechanism relies on the INTEGRITY object within the POLICY_DATA object to guarantee integrity between RSVP Policy Enforcement Points (PEPs) that are not RSVP neighbors. This is useful only when some RSVP nodes are Policy-Ignorant Nodes (PINs). The INTEGRITY object within the POLICY_DATA object MAY be supported by an implementation of this document. Details for the computation of the content of the INTEGRITY object can be found in Appendix B of [RFC2750]. This states that the Policy Decision Point (PDP), at its discretion, and based on destination PEP/PDP or other criteria, selects an Authentication Key and the hash algorithm to be used. Keys to be used between PDPs can be distributed manually or via a standard key management protocol for secure key distribution. Note that where non-RSVP hops may exist in between RSVP hops, as well as where RSVP-capable Policy-Ignorant Nodes (PINs) may exist in between PEPs, it may be difficult for the PDP to determine what the destination PDP is for a POLICY_DATA object contained in some RSVP messages (such as a Path message). This is because in those cases, the next PEP is not known at the time of forwarding the message. In this situation, a key shared across multiple PDPs may be used. This is conceptually similar to the use of a key shared across multiple RSVP neighbors, as discussed in [SEC-GRP-KEY]. We also observe that this issue may not exist in some deployment scenarios where a single
Top   ToC   RFC5946 - Page 32
   PDP (or a low number of PDPs) is used to control all the PEPs of a
   region (such as an administrative domain).  In such scenarios, it may
   be easy for a PDP to determine what the next-hop PDP is, even when
   the next-hop PEP is not known, simply by determining what the next
   region is that will be traversed (say, based on the destination
   address).

6. IANA Considerations

6.1. RSVP Error Codes

Since, as discussed in Section 3.1.2, this document allows two error codes to be used in PathErr messages while [RFC2205] only specified their use in ResvErr messages, IANA has updated the existing entries for these two error codes under the "Error Codes and Globally-Defined Error Value Sub-Codes" registry. Each entry refers to this document, in addition to referring to [RFC2205]. Specifically, the entry for Error Code 1 and Error Code 2 read: 1 Admission Control Failure [RFC2205] [RFC5946] 2 Policy Control Failure [RFC2205] [RFC5946] IANA has also allocated a new RSVP Error Code "36;: Unrecoverable Receiver Proxy Error", as discussed in Section 3.1.2. This error code has been allocated under the "Error Codes and Globally-Defined Error Value Sub-Codes" registry. The entry for this error code reads: 36 Unrecoverable Receiver Proxy Error [RFC5946] The sixteen bits of the Error Value field are defined in [RFC5946]

6.2. Policy Element

This document defines, in Section 4.2, a new policy element called the Receiver Proxy Control policy element. As specified in [RFC2750], standard RSVP policy elements (P-Type values) are to be assigned by IANA as per "IETF Consensus" policy following the policies outlined in [RFC2434] (this policy is now called "IETF Review" as per [RFC5226]). Thus, IANA has allocated one P-Type to the Receiver Proxy Control policy element from the standard RSVP policy element range.
Top   ToC   RFC5946 - Page 33
   In Section 4.2, this document defines a Control-Value field inside
   the Receiver Proxy Control policy element.  IANA has created the
   "Receiver Proxy Control Policy Element (P-Type 0x07) Control-Value
   field" registry and allocated the following values:

   o  0 : Reserved

   o  1 : Receiver-Proxy-Needed

   o  2 : Receiver-Proxy-Not-Needed

   Following the policies outlined in [RFC5226], numbers in the range
   3-127 are allocated according to the "IETF Review" policy, numbers in
   the range 128-240 are assigned on a "First Come First Served" basis,
   and numbers in the range 241-255 are reserved for "Private Use".

7. Acknowledgments

This document benefited from discussions with Carol Iturralde and Anca Zamfir. Lou Berger, Adrian Farrel, and John Drake provided review and guidance, in particular on the usage of the Path_State_Removed flag and of the Notify message, both borrowed from [RFC3473]. We also thank Stephen Kent, Ken Carlberg, and Tim Polk for their valuable input and proposed enhancements. Finally, we thank Cullen Jennings, Magnus Westerlund, and Robert Sparks for stimulating the work on extensions maximizing the reservation span and facilitating migration from the Proxy model to the end-to-end RSVP model.

8. References

8.1. Normative References

[RFC2113] Katz, D., "IP Router Alert Option", RFC 2113, February 1997. [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC2205] Braden, B., Zhang, L., Berson, S., Herzog, S., and S. Jamin, "Resource ReSerVation Protocol (RSVP) -- Version 1 Functional Specification", RFC 2205, September 1997. [RFC2434] Narten, T. and H. Alvestrand, "Guidelines for Writing an IANA Considerations Section in RFCs", BCP 26, RFC 2434, October 1998.
Top   ToC   RFC5946 - Page 34
   [RFC2711]     Partridge, C. and A. Jackson, "IPv6 Router Alert
                 Option", RFC 2711, October 1999.

   [RFC2747]     Baker, F., Lindell, B., and M. Talwar, "RSVP
                 Cryptographic Authentication", RFC 2747, January 2000.

   [RFC2750]     Herzog, S., "RSVP Extensions for Policy Control",
                 RFC 2750, January 2000.

   [RFC3097]     Braden, R. and L. Zhang, "RSVP Cryptographic
                 Authentication -- Updated Message Type Value",
                 RFC 3097, April 2001.

   [RFC3209]     Awduche, D., Berger, L., Gan, D., Li, T., Srinivasan,
                 V., and G. Swallow, "RSVP-TE: Extensions to RSVP for
                 LSP Tunnels", RFC 3209, December 2001.

   [RFC3473]     Berger, L., "Generalized Multi-Protocol Label Switching
                 (GMPLS) Signaling Resource ReserVation Protocol-Traffic
                 Engineering (RSVP-TE) Extensions", RFC 3473,
                 January 2003.

   [RFC4302]     Kent, S., "IP Authentication Header", RFC 4302,
                 December 2005.

   [RFC4303]     Kent, S., "IP Encapsulating Security Payload (ESP)",
                 RFC 4303, December 2005.

   [RFC5226]     Narten, T. and H. Alvestrand, "Guidelines for Writing
                 an IANA Considerations Section in RFCs", BCP 26,
                 RFC 5226, May 2008.

8.2. Informative References

[RFC3181] Herzog, S., "Signaled Preemption Priority Policy Element", RFC 3181, October 2001. [RFC4230] Tschofenig, H. and R. Graveman, "RSVP Security Properties", RFC 4230, December 2005. [RFC5945] Le Faucheur, F., Manner, J., Wing, D., and A. Guillou, "Resource Reservation Protocol (RSVP) Proxy Approaches", RFC 5945, October 2010. [RTR-ALERT] Le Faucheur, F., "IP Router Alert Considerations and Usage", Work in Progress, October 2009.
Top   ToC   RFC5946 - Page 35
   [SEC-GRP-KEY] Behringer, M. and F. Le Faucheur, "Applicability of
                 Keying Methods for RSVP Security", Work in Progress,
                 June 2010.

Authors' Addresses

Francois Le Faucheur Cisco Systems Greenside, 400 Avenue de Roumanille Sophia Antipolis 06410 France Phone: +33 4 97 23 26 19 EMail: flefauch@cisco.com Jukka Manner Aalto University Department of Communications and Networking (Comnet) P.O. Box 13000 FIN-00076 Aalto Finland Phone: +358 9 470 22481 EMail: jukka.manner@tkk.fi URI: http://www.netlab.tkk.fi/~jmanner/ Ashok Narayanan Cisco Systems 300 Beaver Brook Road Boxborough, MA 01719 United States EMail: ashokn@cisco.com Allan Guillou SFR 40-42 Quai du Point du Jour Boulogne-Billancourt 92659 France EMail: allan.guillou@sfr.com Hemant Malik Bharti Airtel, Ltd. 4th Floor, Plot No. 16 Udyog Vihar, Phase IV Gurgaon, 122015 India EMail: Hemant.Malik@airtel.in