<pre>Network Working Group                                           F. Baker
Request for Comments: 4923                                 Cisco Systems
Category: Informational                                          P. Bose
                                                         Lockheed Martin
                                                             August 2007


 <span class="h1">Quality of Service (QoS) Signaling in a Nested Virtual Private Network</span>

Status of This Memo

   This memo provides information for the Internet community.  It does
   not specify an Internet standard of any kind.  Distribution of this
   memo is unlimited.

Copyright Notice

   Copyright (C) The IETF Trust (2007).

Abstract

   Some networks require communication between an interior and exterior
   portion of a Virtual Private Network (VPN) or through a concatenation
   of such networks resulting in a nested VPN, but have sensitivities
   about what information is communicated across the boundary,
   especially while providing quality of service to communications with
   different precedence.  This note seeks to outline the issues and the
   nature of the proposed solutions based on the framework for
   Integrated Services operation over Diffserv networks as described in
   <a href="./rfc2998">RFC 2998</a>.





















<span class="grey">Baker &amp; Bose                 Informational                      [Page 1]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


Table of Contents

   <a href="#section-1">1</a>. Introduction ....................................................<a href="#page-3">3</a>
      <a href="#section-1.1">1.1</a>. Problem Statement ..........................................<a href="#page-3">3</a>
      <a href="#section-1.2">1.2</a>. Background Information and Terminology .....................<a href="#page-4">4</a>
      <a href="#section-1.3">1.3</a>. Nested VPNs ................................................<a href="#page-5">5</a>
      <a href="#section-1.4">1.4</a>. Signaled QoS Technology ....................................<a href="#page-7">7</a>
      <a href="#section-1.5">1.5</a>. The Resource Reservation Protocol (RSVP) ...................<a href="#page-9">9</a>
      <a href="#section-1.6">1.6</a>. Logical Structure of a VPN Router .........................<a href="#page-10">10</a>
   <a href="#section-2">2</a>. Reservation and Preemption in a Nested VPN .....................<a href="#page-13">13</a>
      <a href="#section-2.1">2.1</a>. Reservation in a Nested VPN ...............................<a href="#page-14">14</a>
      <a href="#section-2.2">2.2</a>. Preemption in a Nested VPN ................................<a href="#page-16">16</a>
      <a href="#section-2.3">2.3</a>. Working through an Example ................................<a href="#page-17">17</a>
           2.3.1. Initial Routine Reservations - Generating
                  Network State ......................................<a href="#page-18">18</a>
           2.3.2. Initial Routine Reservations - Request
                  Reservation ........................................<a href="#page-19">19</a>
           <a href="#section-2.3.3">2.3.3</a>. Installation of a Reservation Using Precedence .....<a href="#page-20">20</a>
           <a href="#section-2.3.4">2.3.4</a>. Installation of a Reservation Using Preemption .....<a href="#page-21">21</a>
   <a href="#section-3">3</a>. Data Flows within a VPN Router .................................<a href="#page-24">24</a>
      3.1. VPN Routers That Carry Data across the
           Cryptographic Boundary ....................................<a href="#page-24">24</a>
           <a href="#section-3.1.1">3.1.1</a>. Plaintext to Ciphertext Data Flows .................<a href="#page-24">24</a>
           <a href="#section-3.1.2">3.1.2</a>. Ciphertext to Plaintext Data Flows .................<a href="#page-27">27</a>
      3.2. VPN Routers That Use the Network Guard for
           Signaling across the Cryptographic Boundary ...............<a href="#page-28">28</a>
           <a href="#section-3.2.1">3.2.1</a>. Signaling Flow .....................................<a href="#page-29">29</a>
           <a href="#section-3.2.2">3.2.2</a>. Use Case with Network Guard ........................<a href="#page-30">30</a>
   <a href="#section-4">4</a>. Security Considerations ........................................<a href="#page-33">33</a>
   <a href="#section-5">5</a>. Acknowledgements ...............................................<a href="#page-34">34</a>
   <a href="#section-6">6</a>. References .....................................................<a href="#page-34">34</a>
      <a href="#section-6.1">6.1</a>. Normative References ......................................<a href="#page-34">34</a>
      <a href="#section-6.2">6.2</a>. Informative References ....................................<a href="#page-35">35</a>


















<span class="grey">Baker &amp; Bose                 Informational                      [Page 2]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>.  Introduction</span>

<span class="h3"><a class="selflink" id="section-1.1" href="#section-1.1">1.1</a>.  Problem Statement</span>

   More and more networks wish to guarantee secure transmission of IP
   traffic across public LANs or WANs and therefore use Virtual Private
   Networks.  Some networks require communication between an interior
   and exterior portion of a VPN or through a concatenation of such
   networks resulting in a nested VPN, but have sensitivities about what
   information is communicated across the boundary, especially while
   providing quality of service to communications with different
   precedence.  This note seeks to outline the issues and the nature of
   the proposed solutions.  The outline of the QoS solution for real-
   time traffic has been described at a high level in [<a href="./rfc4542" title="&quot;Implementing an Emergency Telecommunications Service (ETS) for Real-Time Services in the Internet Protocol Suite&quot;">RFC4542</a>].  The
   key characteristics of this proposal are that

   o  it uses standardized protocols,

   o  it includes reservation setup and teardown for guaranteed and
      controlled load services using the standardized protocols,

   o  it is independent of link delay, and therefore consistent with
      high delay*bandwidth networks as well as the more common variety,

   o  it has no single point of failure, such as a central reservation
      manager,

   o  it provides for the preemption of established data flows,

   o  in that preemption, it not only permits a policy-admitted data
      flow in, but selects a specific data flow to exclude based upon
      control input rather than simply accepting a loss of service
      dictated at the discretion of the network control function, and

   o  it interoperates directly with SIP Proxies, H.323 Gatekeepers, or
      other call management subsystems to present the other services
      required in a preemptive or preferential telephone network.

   The thrust of the memo surrounds VPNs that use encryption in some
   form, such as IPsec and their subsequent nesting across multiple
   network domains.  This specific type of VPNs is further clarified in
   <a href="#section-1.3">Section 1.3</a>, which describes the nested VPN as an example of an IPsec
   or IPsec like VPN under the context of a 'customer provisioned' VPN.
   As a result, we will discuss the VPN router supporting "plaintext"
   and "ciphertext" interfaces.  However, the concept extends readily to
   any form of aggregation, including the concept proposed in [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>]
   of the IP traffic entering and leaving a network at identified




<span class="grey">Baker &amp; Bose                 Informational                      [Page 3]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   points, and the use of other kinds of tunnels including Generic
   Routing Encapsulation (GRE), IP/IP, MPLS, and so on.

<span class="h3"><a class="selflink" id="section-1.2" href="#section-1.2">1.2</a>.  Background Information and Terminology</span>

   A note on the use of the words "priority" and "precedence" in this
   document is in order.  The term "priority" has been used in this
   context with a variety of meanings, resulting in a great deal of
   confusion.  The term "priority" is used in this document to identify
   one of several possible datagram scheduling algorithms.  A scheduler
   is used when deciding which datagram will be sent next on a computer
   interface; a priority scheduler always chooses a datagram from the
   highest priority class (queue) that is occupied, shielding one class
   of traffic from most of the jitter by passing jitter it would
   otherwise have experienced to another class.  [<a href="./rfc3181" title="&quot;Signaled Preemption Priority Policy Element&quot;">RFC3181</a>] applies the
   term to a reservation, in a sense that this document will refer to as
   "precedence".  The term "precedence" is used in the sense implied in
   the phrase "Multi-Level Precedence and Preemption" [<a href="#ref-ITU.MLPP.1990" title="&quot;Multilevel Precedence and Preemption Service&quot;">ITU.MLPP.1990</a>];
   some classes of sessions take precedence over others, which may
   result in bandwidth being admitted that might not otherwise have been
   or may result in the prejudicial termination of a lower-precedence
   session under a stated set of circumstances.  For the purposes of the
   present discussion, "priority" is a set of algorithms applied to
   datagrams, where "precedence" is a policy attribute of sessions.  The
   techniques of priority comparisons are used in a router or a policy
   decision point to implement precedence, but they are not the same
   thing.

   Along the same lines, it is important for the reader to understand
   the difference between QoS policies and policies based on the
   "precedence" or "importance" of data to the person or function using
   it.  Voice, regardless of the precedence level of the call, is
   impeded by high levels of variation in network-induced delay.  As a
   result, voice is often serviced using a priority queue, transferring
   jitter from that application's traffic to other applications.  This
   is as true of voice for routine calls as it is for flash traffic.
   There are classes of application traffic that require bounded delay.
   That is a different concept than "no jitter"; they can accept jitter
   within stated bounds.  Routing protocols such as OSPF or BGP are
   critical to the correct functioning of network infrastructure.  While
   they are designed to work well with moderate loss levels, they are
   not helped by them, and even a short period of high loss can result
   in dramatic network events.  Variation in delay, however, is not at
   all an issue if it is within reasonable bounds.  As a result, it is
   common for routers to treat routing protocol datagrams in a way that
   limits the probability of loss, accepting relatively high delay in
   some cases, even though the traffic is absolutely critical to the
   network.  Telephone call setup exchanges have this characteristic as



<span class="grey">Baker &amp; Bose                 Informational                      [Page 4]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   well: faced with a choice between loss and delay, protocols like SIP
   and H.323 far prefer the latter, as the call setup time is far less
   than it would be if datagrams had to be retransmitted, and this is
   true regardless of whether the call is routine or of high precedence.
   As such, QoS markings tell us how to provide good service to an
   application independent of how "important" it may be at the current
   time, while "importance" can be conveyed separately in many cases.

<span class="h3"><a class="selflink" id="section-1.3" href="#section-1.3">1.3</a>.  Nested VPNs</span>

   One could describe a nested VPN network in terms of three network
   diagrams.  Figure 1 shows a simple network stretched across a VPN
   connection.  The VPN router (where, following [<a href="./rfc2460" title="&quot;Internet Protocol, Version 6 (IPv6) Specification&quot;">RFC2460</a>], a "router"
   is "a node that forwards packets not explicitly addressed to
   itself"), performs the following steps:

   o  receives an IP datagram from a plaintext interface,

   o  determines what remote enclave and therefore other VPN router to
      forward it to,

   o  ensures that it has a tunnel mode security association (as
      generally defined in <a href="./rfc4301#section-4">[RFC4301], Section&nbsp;4</a>) with that router,

   o  encloses the encrypted datagram within another VPN (e.g., IPsec)
      and IP header, and

   o  forwards the encapsulated datagram toward the remote VPN router.

      The receiving VPN router reverses the steps:

   o  determines what security association the datagram was received
      from,

   o  decrypts the interior datagram,

   o  forwards the now-decapsulated datagram on a plaintext interface.

   The use of IPsec in this manner is described as the tunnel mode of
   [<a href="./rfc4301" title="&quot;Security Architecture for the Internet Protocol&quot;">RFC4301</a>] and [<a href="./rfc4303" title="&quot;IP Encapsulating Security Payload (ESP)&quot;">RFC4303</a>].











<span class="grey">Baker &amp; Bose                 Informational                      [Page 5]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


           Host  Host  Host       Host  Host  Host
       /------------------/   /------------------/
                 Router -------Router
                            |
                        VPN-Router
                            ||
                            ||   IPsec Tunnel through routed network
                            ||
                        VPN-Router
                            |
                  Router -------Router
       /------------------/   /------------------/
         Host  Host  Host       Host  Host  Host

                      Figure 1: VPN-Connected Enclave

   An important point to understand is that the VPN tunnel, like other
   features of the routed network, are invisible to the host.  The host
   can infer that "something out there" is affecting the Path MTU,
   introducing delay, or otherwise affecting its data stream, but if
   properly implemented, it should be able to adapt to these.  The words
   "if properly implemented" are the bane of every network manager,
   however; substandard implementations do demonstrably exist.

   Outside of the enclave, the hosts are essentially invisible.  The
   communicating enclaves look like a simple data exchange between peer
   hosts across a routed network, as shown in Figure 2.

                                   Hosts Not Visible
                                 /==================/
                                       Router
                                          |
                                     VPN-Router
                               /---------------------/
                                     Inner Domain
                              /---------------------/
                                      VPN-Router
                                          |
                                       Router
                                /==================/
                                 Hosts Not Visible

         Figure 2: VPN-Connected Enclave from Exterior Perspective

   Such networks can be nested and re-used in a complex manner.  As
   shown in Figure 3, a pair of enclaves might communicate across a
   ciphertext network that, for various reasons, is itself re-encrypted
   and transmitted across a larger ciphertext network.  The reasons for



<span class="grey">Baker &amp; Bose                 Informational                      [Page 6]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   doing this vary, but they relate to information-hiding in the wider
   network, different levels of security required for different enclosed
   enclaves, and so on.

             Host  Host  Host       Host  Host  Host
          /------------------/   /------------------/
                     Router -------Router
                               |
                       VPN-Router VPN-Router      VPN-Router
                    /---------------------/    /----------/
                             Router -------------Router
                                        |
                                      VPN-Router      VPN-Router
                                     /-----------/   /----------/
                                          Router -------Router
                                            |
                                            |
                                          Router -------Router
                                     /-----------/   /----------/
                                      VPN-Router      VPN-Router
                                        |
                              Router ------------Router
                    /---------------------/   /----------/
                     VPN-Router VPN-Router     VPN-Router
                               |
                     Router -------Router
          /------------------/   /------------------/
            Host  Host  Host       Host  Host  Host

                           Figure 3: Nested VPN

   The key question this document explores is "how do reservations, and
   preemption of reservations, work in such an environment?"

<span class="h3"><a class="selflink" id="section-1.4" href="#section-1.4">1.4</a>.  Signaled QoS Technology</span>

   The Integrated Services model for networking was originally proposed
   in [<a href="./rfc1633" title="&quot;Integrated Services in the Internet Architecture: an Overview&quot;">RFC1633</a>].  In short, it divides all applications into two broad
   classes: those that will adapt themselves to any available bandwidth,
   and those that will not or cannot.  In the words of [<a href="./rfc1633" title="&quot;Integrated Services in the Internet Architecture: an Overview&quot;">RFC1633</a>]:

        One class of applications needs the data in each packet by a
        certain time and, if the data has not arrived by then, the data
        is essentially worthless; we call these "real-time"
        applications.  Another class of applications will always wait
        for data to arrive; we call these "elastic" applications.





<span class="grey">Baker &amp; Bose                 Informational                      [Page 7]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   The Integrated Services model defines data flows supporting
   applications as either "real-time" or "elastic".  It should be noted
   that "real-time" traffic is also referred to as "inelastic" traffic,
   and that elastic traffic is occasionally referred to as "non-real-
   time".

   In this view, the key issue is the so-called "playback point": a
   real-time application is considered to have a certain point in time
   at which data describing the next sound, picture, or whatever to be
   delivered to a display device or forfeit its utility, while an
   elastic application has no such boundary.  Another way to look at the
   difference is that real-time applications have an irreducible lower
   bound on their bandwidth requirements.  For example, the typical
   G.711 payload is delivered in 160-byte samples (plus 40 bytes of IP/
   UDP/RTP headers) at 20 millisecond intervals.  This will yield 80
   kbps of bandwidth, without silence suppression, and not accounting
   for the layer 2 overhead.  To operate in real-time, a G.711 codec
   requires the network over which its data will be delivered to support
   communications at 80 kbps at the IP layer with roughly constant end-
   to-end delay and nominal or no loss.  If this is not possible (if
   there is significant loss or wide variations in delay), voice quality
   will suffer.  On the other hand, if many megabits of capacity are
   available, the G.711 codec will not increase its bandwidth
   requirements either.  Although adaptive codecs exist (e.g., G.722.2
   or G.726), the adaptive mechanism can either require greater or
   lesser bandwidth and can adapt only within a certain range of
   bandwidth requirements beyond which the quality of the data flow
   required is not met.  Elastic applications, however, will generally
   adapt themselves to any network: if the bottleneck provides 9600 bits
   per second, a Web transfer or electronic mail exchange will happen at
   9600 bits per second, and if hundreds of megabits are available, the
   TCP (or SCTP) transport will increase their transfer rate in an
   attempt to reduce the time required to accomplish the transfer.

   For real-time applications, those that require data to be delivered
   end to end with at least a certain rate and with delays varying
   between stated bounds, the Integrated Services architecture proposes
   the use of a signaling protocol that allows the communicating
   applications and the network to communicate about the application
   requirements and the network's capability to deliver them.  Several
   such protocols have been developed or are under development, notably
   including the Resource Reservation Protocol (RSVP) and Next Steps in
   Signaling (NSIS).  The present discussion is limited to RSVP,
   although any protocol that delivers a similar set of capabilities
   could be considered.






<span class="grey">Baker &amp; Bose                 Informational                      [Page 8]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h3"><a class="selflink" id="section-1.5" href="#section-1.5">1.5</a>.  The Resource Reservation Protocol (RSVP)</span>

   RSVP is initially defined in [<a href="./rfc2205" title="&quot;Resource ReSerVation Protocol (RSVP) -- Version 1 Functional Specification&quot;">RFC2205</a>] with a set of datagram
   processing rules defined in [<a href="./rfc2209" title="&quot;Resource ReSerVation Protocol (RSVP) -- Version 1 Message Processing Rules&quot;">RFC2209</a>] and datagram details for
   Integrated Services [<a href="./rfc2210" title="&quot;The Use of RSVP with IETF Integrated Services&quot;">RFC2210</a>].  Conceptually, this protocol specifies
   a way to identify data flows from a source application to a
   destination application and request specific resources for them.  The
   source may be a single machine or a set of machines listed explicitly
   or implied, whereas the destination may be a single machine or a
   multicast group (and therefore all of the machines in it).  Each
   application is specified by a transport protocol number in the IP
   protocol field, or may additionally include destination and perhaps
   source port numbers.  The protocol is defined for both IPv4 [<a href="./rfc0791" title="&quot;Internet Protocol&quot;">RFC0791</a>]
   and IPv6 [<a href="./rfc2460" title="&quot;Internet Protocol, Version 6 (IPv6) Specification&quot;">RFC2460</a>].  It was recognized immediately that it was also
   necessary to provide a means to perform the same function for various
   kinds of tunnels, which implies a relationship between what is inside
   and what is outside the tunnel.  Definitions were therefore developed
   for IPsec [<a href="./rfc2207" title="&quot;RSVP Extensions for IPSEC Data Flows&quot;">RFC2207</a>] and [<a href="./rfc4860" title="&quot;Generic Aggregate Resource ReSerVation Protocol (RSVP) Reservations&quot;">RFC4860</a>] and for more generic forms of
   tunnels [<a href="./rfc2746" title="&quot;RSVP Operation Over IP Tunnels&quot;">RFC2746</a>].  With the later development of the Differentiated
   Services Architecture [<a href="./rfc2475" title="&quot;An Architecture for Differentiated Services&quot;">RFC2475</a>], definitions were added to specify
   the Differentiated Services Code Point (DSCP) [<a href="./rfc2474" title="&quot;Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers&quot;">RFC2474</a>] to be used by
   a standard RSVP data flow in [<a href="./rfc2996" title="&quot;Format of the RSVP DCLASS Object&quot;">RFC2996</a>] and to use a pair of IP
   addresses and a DSCP as the identifying information for a data flow
   [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>].

   In addition, the initial definition of the protocol included a
   placeholder for policy information, and for preemption of
   reservations.  This placeholder was later specified in detail in
   [<a href="./rfc2750" title="&quot;RSVP Extensions for Policy Control&quot;">RFC2750</a>] with a view to associating a policy [<a href="./rfc2872" title="&quot;Application and Sub Application Identity Policy Element for Use with RSVP&quot;">RFC2872</a>] with an
   identity [<a href="./rfc3182" title="&quot;Identity Representation for RSVP&quot;">RFC3182</a>] and thereby enabling the network to provide a
   contracted service to an authenticated and authorized user.  This was
   integrated with the Session Initiation Protocol [<a href="./rfc3261" title="&quot;SIP: Session Initiation Protocol&quot;">RFC3261</a>] in
   [<a href="./rfc3312" title="&quot;Integration of Resource Management and Session Initiation Protocol (SIP)&quot;">RFC3312</a>].  Preemption of a reservation is specified as in [<a href="./rfc3181" title="&quot;Signaled Preemption Priority Policy Element&quot;">RFC3181</a>]
   -- a reservation that is installed in the network using an Preemption
   Priority and retained using a separate Defending Priority may be
   removed by the network via an RESV Error signal that removes the
   entire reservation.  This has issues, however, in that the matter is
   often not quite so black and white.  If the issue is that an existing
   reservation for 80 kbps can no longer be sustained but a 60 kbps
   reservation could, it is possible that a VoIP sender could change
   from a G.711 codec to a G.729 codec and achieve that.  Or, if there
   are multiple sessions in a tunnel or other aggregate, one of the
   calls could be eliminated leaving capacity for the others.  [<a href="./rfc4495" title="&quot;A Resource Reservation Protocol (RSVP) Extension for the Reduction of Bandwidth of a Reservation Flow&quot;">RFC4495</a>]
   seeks to address this issue.

   In a similar way, a capability was added to limit the possibility of
   control signals being spoofed or otherwise attacked [<a href="./rfc2747" title="&quot;RSVP Cryptographic Authentication&quot;">RFC2747</a>]
   [<a href="./rfc3097" title="&quot;RSVP Cryptographic Authentication -- Updated Message Type Value&quot;">RFC3097</a>].



<span class="grey">Baker &amp; Bose                 Informational                      [Page 9]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   [<a id="ref-RFC3175">RFC3175</a>] describes several features that are unusual in RSVP, being
   specifically set up to handle aggregates in a service provider
   network.  It describes three key components:

   o  The <a href="./rfc3175">RFC 3175</a> session object, which identifies not the IP addresses
      of the packets that are identified, but the IP addresses of the
      ingress and egress devices in the network, and the DSCP that the
      traffic will use.

   o  The function of a reservation "aggregator", which operates in the
      ingress router and accepts individual reservations from the
      "customer" network.  It aggregates the reservations into the ISP
      core in a tunnel or an MPLS LSP, or as a traffic stream that is
      known to leave at the deaggregator.

   o  The function of a reservation "deaggregator", which operates in
      the egress router and breaks the aggregate reservation and data
      streams back out into individual data streams that may be passed
      to other networks.

   In retrospect, the Session Object specified by <a href="./rfc3175">RFC 3175</a> is useful but
   not intrinsically necessary.  If the ISP network uses tunnels, such
   as MPLS LSPs, IP/IP or GRE tunnels or enclosing IPsec Security
   Associations, the concepts of an aggregator and a deaggregator work
   in the same manner, although the reservation mechanism would be that
   of [<a href="./rfc3473" title="&quot;Generalized Multi-Protocol Label Switching (GMPLS) Signaling Resource ReserVation Protocol-Traffic Engineering (RSVP-TE) Extensions&quot;">RFC3473</a>] and [<a href="./rfc3474" title="&quot;Documentation of IANA assignments for Generalized MultiProtocol Label Switching (GMPLS) Resource Reservation Protocol - Traffic Engineering (RSVP-TE) Usage and Extensions for Automatically Switched Optical Network (ASON)&quot;">RFC3474</a>], [<a href="./rfc2207" title="&quot;RSVP Extensions for IPSEC Data Flows&quot;">RFC2207</a>], [<a href="./rfc4860" title="&quot;Generic Aggregate Resource ReSerVation Protocol (RSVP) Reservations&quot;">RFC4860</a>], or [<a href="./rfc2746" title="&quot;RSVP Operation Over IP Tunnels&quot;">RFC2746</a>].

<span class="h3"><a class="selflink" id="section-1.6" href="#section-1.6">1.6</a>.  Logical Structure of a VPN Router</span>

   The conceptual structure of a VPN router is similar to that of any
   other router.  In its simplest form, it is physically a two or more
   port device (similar to that shown in Figure 4), which has one or
   more interfaces to the protected enclave(s) and one or more
   interfaces to the outside world.  On the latter, it structures some
   number of tunnels (in the case of an IPsec tunnel, having security
   associations) that it can treat as point-to-point interfaces from a
   routing perspective.














<span class="grey">Baker &amp; Bose                 Informational                     [Page 10]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


          +---------+  +-------+   +----+----+       +---------+
          |   RSVP  |  |Routing|   |Net Guard|        |IPsec Mgr|
          +----+----+  +---+---+   +----+----+       +----+----+
               |           |            |                 |
          +----+-----------+------------+-----------------+----+
          |                         IP                         |
          +-----------+--------------------+------------+------+
                      |                    |            |
                      |              +-----+-----+ +----+------+
                      |              | Encrypt/  | | Encrypt/  |
                      |              |Decrypt for| |Decrypt for|
                      |              | Security  | | Security  |
                      |              |Association| |Association|
                      |              +-----+-----+ +----+------+
                      |                    |            |
          +-----------+------------+ +-----+------------+------+
          |       Plaintext        | |       Ciphertext        |
          |       Interface        | |       Interface         |
          +------------------------+ +-------------------------+

                Figure 4: Logical Structure of a VPN Router

   The encrypt/decrypt unit may be implemented as a function of the
   plaintext router, as a function on its interface card, or as a
   function of an external device with a private interface to the IP
   routing functionality of the plaintext router.  These are
   conceptually equivalent, although there are practical differences in
   implementation.  The key issue is that when IP routing presents a
   message to the encrypt/decrypt unit for transmission, it must also be
   presented with the IP address of the plaintext routing peer, whether
   host or router, to which the security association must be
   established.  This IP Address is used to select (and perhaps create)
   the security association, and in turn select the appropriate set of
   security parameters.  This could also be implemented by presenting
   the local Security Parameter Index (SPI) for the data, if it has been
   created out of band by the Network Management Process.

   In addition, it is necessary for aggregated signaling to be generated
   for the ciphertext domain.  This may be accomplished in several ways:

   o  by having the RSVP process on the plaintext router generate the
      messages and having the encrypt/decrypt unit bypass them into the
      ciphertext network

   o  by having the plaintext RSVP process advise a process in the
      encrypt/decrypt implementation of what needs to be generated using
      some local exchange, and having it generate such messages, or




<span class="grey">Baker &amp; Bose                 Informational                     [Page 11]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   o  by having a separate parallel network management system
      intermediate between the plaintext and ciphertext routers, in
      which case, the encrypt/decrypt unit and the parallel network
      system must use the same address, and the ciphertext router must
      distinguish between traffic for them based on SPI or the presence
      of encryption.

   Control plane signaling using this additional path is described in
   <a href="#section-3.2">Section 3.2</a>.  The information flow between the plaintext and
   ciphertext domains includes

   o  IP datagrams via the encrypt/decrypt unit,

   o  RSVP signaling via the bypass path,

   o  Control information coordinating security associations, and

   o  precious little else.

































<span class="grey">Baker &amp; Bose                 Informational                     [Page 12]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>.  Reservation and Preemption in a Nested VPN</span>

                        /                           \
                       (       +--+   +--+   enclave )   ,---------.
         .----------.   \      |H2+---+R2|          / ,-'           `
          +--+   +--+`--.\     +--+   ++-+         / /   +--+   +--+
          |H1+---+R1|    \`.           |         ,' /    |R3+---+H3|
          +--+   +-++     ) '--.    +----++  _.-'  (     ++-+   +--+
                   |     /    _.`---|VPN2||''-.     \     |
         enclave +----+-i.--''      +----++    `----.\ +----+ enclave
         --------|VPN1|'              |              ``|VPN3|       ,
                ,+----+               |                +----+------'
              ,' --+-------+----------+------------------+---`.
            ,'            ++-+                                 `.
          ,'              |R7+--------+                          `.
         / interface      +--+        |                            \
           domain 1                 +-+--+                          \
                          _.--------|VPN7|--------.
                  ,-----''          +--+-+         `------.         .
         `-.   ,-'                     |                   `-.   .-'
            `-:  inner domain        +-++                     `.'
            (                        |R9|                       )
            .'.                      ++-+                     ;-.
          .'   `-.                    |                    ,-'   `-.
         '        `------.          +-+--+         _.-----'         `
           interface      `---------|VPN8|-------''
           domain 2                 +-+--+                          /
         \                            |          +--+              /
          `.                          +----------+R8|            ,'
            `.                                   ++-+          ,'
              `. --+------------------+-----------+------+-- ,'
           ,-----+----+               |                +----+------.
         ,'      |VPN6|.              |              _.|VPN4|       `
                 +----+.`----.      +----+     _.--'' ,+----+
                  |     \     `--=.-|VPN5|---:'      /    |
          +--+   ++-+    :   ,-''   +----+    `--.  ;    ++-+   +--+
          |H6+---+R6|    | ,'          |          `.|    |R4+---+H4|
          +--+   +--+    ;/    +--+   ++-+          :    +--+   +--+
                        //     |H5+---+R5|           \
          enclave     ,'(      +--+   +--+            `.     enclave
         `.         ,'   \                 enclave   /  '-.         ,
           `-------'      \                         /      `-------'

                  Figure 5: Reservations in a Nested VPN

   Let us discuss how a resource reservation protocol, and specifically
   RSVP, might be used in a nested virtual private network.




<span class="grey">Baker &amp; Bose                 Informational                     [Page 13]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h3"><a class="selflink" id="section-2.1" href="#section-2.1">2.1</a>.  Reservation in a Nested VPN</span>

   A reservation in a nested VPN is very much like a reservation in any
   other network, with one exception: it is composed of multiple
   reservations that must be coordinated.  These include a reservation
   within the originating and receiving enclaves and a reservation at
   each layer of the VPN, as shown in Figure 5.

   Thus, when a host in one enclave opens a reservation to a host in
   another enclave, a reservation of the appropriate type and size is
   created end to end.  As it traverses the VPN router leaving its
   enclave, the reservation information and the data are placed within
   the appropriate tunnel (e.g., the IPsec Security Association for its
   precedence level to the appropriate remote VPN router).  At the
   remote VPN router, it is extracted from the tunnel and passed on its
   way to the target system.  The data in the enclave will be marked
   with a DSCP appropriate to its application and (if there is a
   difference) precedence level, and the signaling datagrams (PATH and
   RESV) are marked with a DCLASS object indicating that value.  RSVP
   signaling datagrams (PATH and RESV) are marked with a DCLASS object
   indicating the value used for the corresponding data.  The DSCP on
   the signaling datagrams, however, is a DSCP for signaling, and has
   the one provision that if routing varies by DSCP, then it must be a
   DSCP that is routed the same way as the relevant data.  The [<a href="./rfc2872" title="&quot;Application and Sub Application Identity Policy Element for Use with RSVP&quot;">RFC2872</a>]
   policy object specifies the applicable policy (e.g., "routine service
   for voice traffic") and asserts a [<a href="./rfc3182" title="&quot;Identity Representation for RSVP&quot;">RFC3182</a>] credential indicating its
   user (which may be a person or a class of persons).  As specified in
   [<a href="./rfc3181" title="&quot;Signaled Preemption Priority Policy Element&quot;">RFC3181</a>], it also specifies its Preemption Priority and its
   Defending Priority; these enable the Preemption Priority of a new
   session to be compared with the Defending Priority of previously
   admitted sessions.

   On the ciphertext side of the VPN router, no guarantees result unless
   the VPN router likewise sets up a reservation to the peer VPN Router
   across the ciphertext domain.  Thus, the VPN router sets up an
   [<a href="./rfc2207" title="&quot;RSVP Extensions for IPSEC Data Flows&quot;">RFC2207</a>], [<a href="./rfc4860" title="&quot;Generic Aggregate Resource ReSerVation Protocol (RSVP) Reservations&quot;">RFC4860</a>], or [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] reservation to its peer.

   The Session Object defined by [<a href="./rfc2207" title="&quot;RSVP Extensions for IPSEC Data Flows&quot;">RFC2207</a>] or [<a href="./rfc4860" title="&quot;Generic Aggregate Resource ReSerVation Protocol (RSVP) Reservations&quot;">RFC4860</a>] contains a field
   called a "virtual destination port", which allows the multiplexing of
   many reservations over a common security association and, in the
   latter case, a common DSCP value.  Thus, the voice traffic at every
   precedence level might use the Expedited Forwarding (EF) DSCP and
   service as described in [<a href="./rfc3246" title="&quot;An Expedited Forwarding PHB (Per-Hop Behavior)&quot;">RFC3246</a>], but the reservations would be for
   "the aggregate of voice sessions at precedence Pn between these VPN
   routers".  This would allow the network administration to describe
   policies with multiple thresholds, such as "a new session at
   precedence Pn may be accepted if the total reserved bandwidth does
   not exceed threshold Qn; if it is necessary and sufficient to accept



<span class="grey">Baker &amp; Bose                 Informational                     [Page 14]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   the reservation, existing reservations at lower precedences may be
   preemptively reduced to make acceptance of the new session possible".

   In the [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] case, since the DSCP must be used to identify both
   the reservation and the corresponding data stream, the aggregate
   reservations for different precedence levels require different DSCP
   values.

   In either case, the fundamental necessity is for one VPN router to
   act as what [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] calls the "aggregator" and another to act as
   the "deaggregator", and extend a VPN tunnel between them.  If the VPN
   Tunnel is an IPsec Security Association between the VPN routers and
   the IP packet is entirely contained within (such as is used to cross
   a firewall), then the behavior of [<a href="./rfc2746" title="&quot;RSVP Operation Over IP Tunnels&quot;">RFC2746</a>] is required of the
   tunnel.  That bearer will have the following characteristics:

   o  it will have a DSCP corollary to the DSCP for the data or the same
      DSCP as the data it carries,

   o  the reservations and data will be carried in security associations
      between the VPN routers, and

   o  the specification for the reservation for the tunnel itself will
      not be less than the sum of the requirements of the aggregated
      reservations.

   The following requirements relationships apply between the set of
   enclosed reservations and the tunnel reservation:

   o  The sum of the average rates of the contained reservations, having
      been adjusted for the additional IP headers, will be less than or
      equal to the average rate of the tunnel reservation.

   o  The sum of the peak rates of the contained reservations, having
      been adjusted for the additional IP headers, will be less than or
      equal to the peak rate of the tunnel reservation.

   o  The sum of the burst sizes of the contained reservations, having
      been adjusted for the additional IP headers, will be less than or
      equal to the burst size of the tunnel reservation.

   o  The Preemption Priority of a tunnel reservation is identical to
      that of the individual reservations it aggregates.

   o  The Defending Priority of a tunnel reservation is identical to
      that of the individual reservations it aggregates.





<span class="grey">Baker &amp; Bose                 Informational                     [Page 15]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   This would differ only in the case that measurement-based admission
   is in use in the tunnel but not in the end system.  In that case, the
   tunnel's average bandwidth specification would be greater than or
   equal to the actual average offered traffic.  Such systems are beyond
   the scope of this specification.

   As a policy matter, it may be useful to note a quirk in the way
   Internet QoS works.  If the policies for various precedence levels
   specify different thresholds (e.g., "to accept a new routine call,
   the total reserved bandwidth after admission may not exceed X; to
   accept a call with a higher precedence level, the total reserved
   bandwidth after admission may not exceed X+Y, and this may be
   achieved by preempting a call with a lower precedence level"), the
   bandwidth Y effectively comes from the bandwidth in use by elastic
   traffic rather than forcing a preemption event.

<span class="h3"><a class="selflink" id="section-2.2" href="#section-2.2">2.2</a>.  Preemption in a Nested VPN</span>

   As discussed in <a href="#section-1.5">Section 1.5</a>, preemption is specified in [<a href="./rfc3181" title="&quot;Signaled Preemption Priority Policy Element&quot;">RFC3181</a>] and
   further addressed in [<a href="./rfc4495" title="&quot;A Resource Reservation Protocol (RSVP) Extension for the Reduction of Bandwidth of a Reservation Flow&quot;">RFC4495</a>].  The issue is that in many cases the
   need is to reduce the bandwidth of a reservation due to a change in
   the network, not simply to remove the reservation.  In the case of an
   end-system-originated reservation, the end system might be able to
   accommodate the need through a change of codec; in the case of an
   aggregate of some kind, it could reduce the bandwidth it is sending
   by dropping one or more reservations entirely.

   In a nested VPN or other kind of aggregated reservation, this means
   that the deaggregator (the VPN router initiating the RESV signal for
   the tunnel) must

   o  receive the RESV Error signaling it to reduce its bandwidth,

   o  re-issue its RESV accordingly,

   o  identify one or more of its aggregated reservations, enough to do
      the job, and

   o  signal them to reduce their bandwidth accordingly.

   It is possible, of course, that it is signaling them to reduce their
   bandwidth to zero, which is functionally equivalent to removing the
   reservation as described in [<a href="./rfc3181" title="&quot;Signaled Preemption Priority Policy Element&quot;">RFC3181</a>].

   In the routers in the core, an additional case arises.  One could
   imagine that some enclave presents the VPN with a single session, and
   that session has a higher precedence level.  If some interior link is
   congested (e.g., the reserved bandwidth will exceed policy if the



<span class="grey">Baker &amp; Bose                 Informational                     [Page 16]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   call is admitted), a session between a different pair of VPN routers
   must be preempted.  More generally, in selecting a reservation to
   preempt, the core router must always select a reservation at the
   lowest available Defending Priority.  This is the reason that various
   precedence levels must be kept separate in the core.

<span class="h3"><a class="selflink" id="section-2.3" href="#section-2.3">2.3</a>.  Working through an Example</span>

   The network in Figure 5 shows three security layers: six plaintext
   enclaves around the periphery, two ciphertext domains connecting them
   at one layer (referred to in the diagram as an "interface domain"),
   and a third ciphertext domain connecting the first two (referred to
   in the diagram as an "inner domain").  The following distribution of
   information exists:

   o  Each enclave has access to general routing information concerning
      other enclaves it is authorized to communicate with: systems in it
      can translate a DNS name for a remote host or domain and obtain
      the corresponding address or prefix.

   o  Each enclave router also has specific routing information
      regarding its own enclave.

   o  A default route is distributed within the enclave, pointing to its
      VPN router.

   o  VPN Routers 1-6 are able to translate remote enclave prefixes to
      the appropriate remote enclave's VPN router addresses.

   o  Each interface domain has access to general routing information
      concerning the other interface domains, but not the enclaves.
      Systems in an interface domain can translate a DNS name for a
      remote interface domain and obtain the corresponding address or
      prefix.

   o  Each interface domain router also has specific routing information
      regarding its own interface domain.

   o  A default route is distributed within the interface domain,
      pointing to the "inner" VPN router.

   o  VPN Routers 7 and 8 are able to translate remote interface domain
      prefixes to remote VPN router addresses.

   o  Routers in the inner domain have routing information for that
      domain only.





<span class="grey">Baker &amp; Bose                 Informational                     [Page 17]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   While the example shows three levels, there is nothing magic about
   the number three.  The model can be extended to any number of
   concentric layers.

   Note that this example places unidirectional reservations in the
   forward direction.  In voice and video applications, one generally
   has a reservation in each direction.  The reverse direction is not
   discussed, for the sake of clarity, but operates in the same way in
   the reverse direction and uses the same security associations.

<span class="h4"><a class="selflink" id="section-2.3.1" href="#section-2.3.1">2.3.1</a>.  Initial Routine Reservations - Generating Network State</span>

   Now let us install a set of reservations from H1 to H4, H2 to H5, and
   H3 to H6, and for the sake of argument, let us presume that these are
   at the "routine" precedence.  H1, H2, and H3 each initiate a PATH
   signal describing their traffic.  For the sake of argument, let us
   presume that H1's reservation is for an [<a href="./rfc2205" title="&quot;Resource ReSerVation Protocol (RSVP) -- Version 1 Functional Specification&quot;">RFC2205</a>] session, H2's
   reservation is for a session encrypted using IPsec, and therefore
   depends on [<a href="./rfc2207" title="&quot;RSVP Extensions for IPSEC Data Flows&quot;">RFC2207</a>], and H3 (which is a Public Switched Telephone
   Network (PSTN) gateway) sends an [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] reservation comprising a
   number of distinct sessions.  Since these are going to H4, H5, and
   H6, respectively, the default route leads them to VPN1, VPN2, and
   VPN3, respectively.

   The VPN routers each ensure that they have an appropriate security
   association or tunnel open to the indicated remote VPN router (VPN4,
   VPN5, or VPN6).  This will be a security association or tunnel for
   the indicated application at the indicated precedence level.  Having
   accomplished that, it will place the PATH signal into the security
   association and forward it.  If such does not already exist,
   following [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>]'s aggregation model, it will now open a
   reservation (send a PATH signal) for the tunnel/SA within the
   interface domain; if the reservation does exist, the VPN router will
   increase the bandwidth indicated in the ADSPEC appropriately.  In
   this example, these tunnel/SA reservations will follow the default
   route to VPN7.

   VPN7 ensures that it has an appropriate security association or
   tunnel open to VPN8.  This will be a security association or tunnel
   for the indicated application at the indicated precedence level.
   Having accomplished that, it will place the PATH signal into the
   security association and forward it.  If such does not already exist,
   following [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>]'s aggregation model, it will now open a
   reservation (send a PATH signal) for the tunnel/SA within the
   interface domain; if the reservation does exist, the VPN router will
   increase the bandwidth indicated in the ADSPEC appropriately.  In
   this example, this tunnel/SA reservation is forwarded to VPN8.




<span class="grey">Baker &amp; Bose                 Informational                     [Page 18]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   VPN8 acts as an [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] deaggregator for the inner domain.  This
   means that it receives the PATH signal for the inner domain
   reservation and stores state, decrypts the data stream from VPN7,
   operates on the RSVP signals as an RSVP-configured router, and
   forwards the received IP datagrams (including the updated PATH
   signals) into its interface domain.  The PATH signals originated by
   VPN1, VPN2, and VPN3 are therefore forwarded towards VPN4, VPN5, and
   VPN6 according to the routing of the interface domain.

   VPN4, VPN5, and VPN6 each act as an [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] deaggregator for the
   interface domain.  This means that it receives the PATH signal for
   the interface domain reservation and stores state, decrypts the data
   stream from its peer, operates on the RSVP signals as an RSVP-
   configured router, and forwards the received IP datagrams (including
   the updated PATH signals) into its enclave.  The PATH signals
   originated by H1, H2, and H3 are therefore forwarded towards H4, H5,
   and H6 according to the routing of the enclave.

   H4, H5, and H6 now receive the original PATH signals and deliver them
   to their application.

<span class="h4"><a class="selflink" id="section-2.3.2" href="#section-2.3.2">2.3.2</a>.  Initial Routine Reservations - Request Reservation</span>

   The application in H4, H5, and H6 decides to install the indicated
   reservations, meaning that they now reply with RESV signals.  These
   signals request the bandwidth reservation.  Following the trail left
   by the PATH signals, the RESV signals traipse back to their
   respective sources.  The state left by the PATH signals leads them to
   VPN4, VPN5, and VPN6, respectively.  If the routers in the enclaves
   are configured for RSVP, this will be explicitly via R4, R5, or R6;
   if they are not, routing will lead them through those routers.

   The various RSVP-configured routers en route in the enclave
   (including the VPN router on the "enclave" side) will verify that
   there is sufficient bandwidth on their links and that any other
   stated policy is also met.  Having accomplished that, each will
   update its reservation state and forward the RESV signal to the next.
   The VPN routers will also each generate an RESV for the reservation
   within the interface domain, attempting to set or increase the
   bandwidth of the reservation appropriately.

   The various RSVP-configured routers en route in the interface domain
   (including VPN8) will verify that there is sufficient bandwidth on
   their links and that any other stated policy is also met.  Having
   accomplished that, each will update its reservation state and forward
   the RESV signal to the next.  VPN8 will also generate an RESV for the





<span class="grey">Baker &amp; Bose                 Informational                     [Page 19]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   reservation within the inner domain, attempting to set or increase
   the bandwidth of the reservation appropriately.  This gets the
   reservation to the inner deaggregator, VPN8.

   The various RSVP-configured routers en route in the inner domain
   (including VPN7) will verify that there is sufficient bandwidth on
   their links and that any other stated policy is also met.  Having
   accomplished that, each will update its reservation state and forward
   the RESV signal to the next.  This gets the signal to VPN7.

   VPN7 acts as an [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] aggregator for the inner domain.  This
   means that it receives the RESV signal for the inner domain
   reservation and stores state, decrypts the data stream from VPN8,
   operates on the RSVP signals as an RSVP-configured router, and
   forwards the received IP datagrams (including the updated RESV
   signals) into its interface domain.  The RESV signals originated by
   VPN4, VPN5, and VPN6 are therefore forwarded towards VPN1, VPN2, and
   VPN3 through the interface domain.

   VPN1, VPN2, and VPN3 each act as an [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] aggregator for the
   interface domain.  This means that it receives the RESV signal for
   the interface domain reservation and stores state, decrypts the data
   stream from its peer, operates on the RSVP signals as an RSVP-
   configured router, and forwards the received IP datagrams (including
   the updated RESV signals) into its enclave.  The RESV signals
   originated by H4, H5, and H6 are therefore forwarded towards H1, H2,
   and H3 according to the routing of the enclave.

   H1, H2, and H3 now receive the original RESV signals and deliver them
   to their application.

<span class="h4"><a class="selflink" id="section-2.3.3" href="#section-2.3.3">2.3.3</a>.  Installation of a Reservation Using Precedence</span>

   Without going through the details called out in Sections <a href="#section-2.3.1">2.3.1</a> and
   2.3.2, if sufficient bandwidth exists to support them, reservations
   of other precedence levels or other applications may also be
   installed across this network.  If the "routine" reservations already
   described are for voice, for example, and sufficient bandwidth is
   available under the relevant policy, a reservation for voice at the
   "priority" precedence level might be installed.  Due to the mechanics
   of preemption, however, this would not expand the existing "routine"
   reservations in the interface and inner domains, as doing this causes
   loss of information - how much of the reservation is now "routine"
   and how much is "priority"?  Rather, this new reservation will open
   up a separate set of tunnels or security associations for traffic of
   its application class at its precedence between that aggregator and
   deaggregator.




<span class="grey">Baker &amp; Bose                 Informational                     [Page 20]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-21" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   As a side note, there is an opportunity here that does not exist in
   the PSTN.  In the PSTN, all circuits are potentially usable by any
   PSTN application under a certain set of rules (H channels, such as
   those used by video streams, must be contiguous and ordered).  As
   such, if a channel is not made available to routine traffic but is
   made available to priority traffic, the operator is potentially
   losing revenue on the reserved bandwidth and deserves remuneration.
   However, in the IP Internet, some bandwidth must be kept for basic
   functions such as routing, and, in general, policies will not permit
   100% of the bandwidth on an interface to be allocated to one
   application at one precedence.  As a result, it may be acceptable to
   permit a certain portion (e.g., 50%) to be used by routine voice and
   a larger amount (e.g., 60%) to be used by voice at a higher
   precedence level.  Under such a policy, a higher precedence
   reservation for voice might not result in the preemption of a routine
   call, but rather impact elastic traffic, and might be accepted at a
   time that a new reservation of lower precedence might be denied.

   In microwave networks, such as satellite or mobile ad hoc, one could
   also imagine network management intervention that could change the
   characteristics of the radio signal to increase the bandwidth under
   some appropriate policy.

<span class="h4"><a class="selflink" id="section-2.3.4" href="#section-2.3.4">2.3.4</a>.  Installation of a Reservation Using Preemption</span>

   So we now have a number of reservations across the network described
   in Figure 5 including several reservations at "routine" precedence
   and one at "priority" precedence.  For sake of argument, let us
   presume that the link from VPN7 to R9 is now fully utilized - all of
   the bandwidth allocated by policy to voice at the routine or priority
   level has been reserved.  Let us further imagine that a new
   "priority" reservation is now placed from H3 to H6.

   The process described in <a href="#section-2.3.1">Section 2.3.1</a> is followed, resulting in PATH
   state across the network for the new reservation.  This is installed
   even though it is not possible to install a new reservation on VPN7-
   R9, as it does not install any reservation and the network does not
   know whether H6 will ultimately require a reservation.

   The process described in <a href="#section-2.3.2">Section 2.3.2</a> is also followed.  The
   application in H6 decides to install the indicated reservation,
   meaning that it now replies with an RESV signal.  Following the trail
   left by the PATH signal, the RESV signal traipses back towards H3.
   VPN6 and (if RSVP was configured) R6 verify that there is sufficient
   bandwidth on their links and that any other stated policy is also
   met.  Having accomplished that, each will update its reservation





<span class="grey">Baker &amp; Bose                 Informational                     [Page 21]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-22" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   state and forward the RESV signal to the next.  VPN6 also generates
   an RESV for the reservation within the interface domain, attempting
   to set or increase the bandwidth of the reservation appropriately.

   VPN6, R8, and VPN8's "interface domain" sides now verify that there
   is sufficient bandwidth on their links and that any other stated
   policy is also met.  Having accomplished that, each will update its
   reservation state and forward the RESV signal to the next.  VPN8 will
   also generate an RESV for the reservation within the inner domain,
   attempting to set or increase the bandwidth of the reservation
   appropriately.  This gets the reservation to the inner deaggregator,
   VPN8.

   VPN8's "inner domain" side and R9 now verify that there is sufficient
   bandwidth on their links and that any other stated policy is also
   met.  At R9, a problem is detected - there is not sufficient
   bandwidth under the relevant policy.  In the absence of precedence,
   R9 would now return an RESV Error indicating that the reservation
   could not be increased or installed.  In such a case, if a
   preexisting reservation of lower bandwidth already existed, the
   previous reservation would remain in place but the new bandwidth
   would not be granted, and the originator (H6) would be informed.  Let
   us clarify what it means to be at a stated precedence: it means that
   the POLICY_DATA object in the RESV contains a Preemption Priority and
   a Defending Priority with values specified in some memo.  With
   precedence, [<a href="./rfc4495" title="&quot;A Resource Reservation Protocol (RSVP) Extension for the Reduction of Bandwidth of a Reservation Flow&quot;">RFC4495</a>]'s algorithm would have the Preemption Priority
   of the new reservation compared to the Defending Priority of extant
   reservations in the router, of which there are two: one VPN7-&gt;VPN8 at
   "routine" precedence and one VPN7-&gt;VPN8 at "priority" precedence.
   Since the Defending Priority of routine reservation is less than the
   Preemption Priority of a "priority" reservation, the "routine"
   reservation is selected.  R9 determines that it will accept the
   increase in its "priority" reservation VPN7-&gt;VPN8 and reduce the
   corresponding "routine" reservation.  Two processes now occur in
   parallel:

   o  The routine reservation is reduced following the algorithms in
      [<a href="./rfc4495" title="&quot;A Resource Reservation Protocol (RSVP) Extension for the Reduction of Bandwidth of a Reservation Flow&quot;">RFC4495</a>] and

   o  The priority reservation continues according to the usual rules.

   R9 reduces its "routine" reservation by sending an RESV Error
   updating its internal state to reflect the reduced reservation and
   sending an RESV Error to VPN8 requesting that it reduce its
   reservation to a number less than or equal to the relevant threshold
   less the sum of the competing reservations.  VPN8, acting as a
   deaggregator, makes two changes.  On the "inner domain" side, it
   marks its reservation down to the indicated rate (the most it is now



<span class="grey">Baker &amp; Bose                 Informational                     [Page 22]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-23" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   permitted to reserve), so that if an RESV Refresh event happens, it
   will request the specified rate.  On the "interface domain" side, it
   selects one or more of the relevant reservations by an algorithm of
   its choosing and requests that it likewise reduce its rate.  For the
   sake of argument, let us imagine that the selected reservation is the
   one to VPN5.  The RESV Error now makes its way through R8 to VPN5,
   which similarly reduces its bandwidth request to the stated amount
   and passes a RESV Error signal on the "enclave" side requesting that
   the reservation be appropriately reduced.

   H5 is now faced with a decision.  If the request is to reduce its
   reservation to zero, that is equivalent to tearing down the
   reservation.  In this simple case, it sends an RESV Tear to tear down
   the reservation entirely and advises its application to adjust its
   expectations of the session accordingly, which may mean shutting down
   the session.  If the request is to reduce it below a certain value,
   however, it may be possible for the application to do so and remain
   viable.  For example, if a VoIP application using a G.711 codec (80
   kbps) is asked to reduce its bandwidth below 70 kbps, it may be
   possible to renegotiate the codec in use to G.729 or some other
   codec.  In such a case, the originating application should re-reserve
   at the stated bandwidth (in this case, 70 kbps), initiate the
   application level change, and let the application change the
   reservation again (perhaps to 60 kbps) when it has completed that
   process.

   At the time the reservation is being processed at R9, for the
   "priority" reservation, R9 believes that it has sufficient bandwidth
   and that any other stated policy is also met, and it forwards the
   RESV to VPN7.  Each will update its reservation state and forward the
   RESV signal to the next.  VPN7 now acts as an [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] aggregator
   for the inner domain.  This means that it receives the RESV signal
   for the inner domain reservation and stores state, decrypts the data
   stream from VPN8, operates on the RSVP signals as an RSVP-configured
   router, and forwards the received IP datagrams (including the updated
   RESV signals) into its interface domain.  The RESV signals originated
   by VPN4, VPN5, and VPN6 are therefore forwarded towards VPN1, VPN2,
   and VPN3 through the interface domain.

   VPN3 now acts as an [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] aggregator for the interface domain.
   This means that it receives the RESV signal for the interface domain
   reservation and stores state, decrypts the data stream from its peer,
   operates on the RSVP signals as an RSVP-configured router, and
   forwards the received IP datagrams (including the updated RESV
   signals) into its enclave.  The RESV signal originated by H6 is
   therefore forwarded towards H3 according to the routing of the
   enclave.




<span class="grey">Baker &amp; Bose                 Informational                     [Page 23]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-24" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   H3 now receives the original RESV signals and delivers it to the
   relevant application.

<span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>.  Data Flows within a VPN Router</span>

   This section details the data flows within a VPN router, in the
   context of sessions as described in <a href="#section-2">Section 2</a>.  It specifically
   identifies the signaling flow at a given VPN boundary and
   additionally elaborates the signaling mechanism with the aid of a
   Network Guard.  A use case describing the proposal in the context of
   an operational scenario is presented herein.

<span class="h3"><a class="selflink" id="section-3.1" href="#section-3.1">3.1</a>.  VPN Routers That Carry Data across the Cryptographic Boundary</span>

<span class="h4"><a class="selflink" id="section-3.1.1" href="#section-3.1.1">3.1.1</a>.  Plaintext to Ciphertext Data Flows</span>

          +-----------------------+    +----------------------+
          | +--------------------+|    |+--------------------+|
          | |RSVP                ||    ||Aggregate RSVP      ||
          | |                    ||    ||                    ||
          | |Per session:        || ID ||Agg. Session        ||
          | |  Destination       ||---&gt;||  Agg. Destination  ||
          | |  Source            ||    ||  Agg. Source= self ||
          | |  potential SPI     ||    ||  Agg. SPI generated||
          | |  DSCP             ---------&gt; DSCP              ||
          | |  vPort or protocol---------&gt; vPort             ||
          | |           and port ||    ||                    ||
          | |  Mean rate        ---------&gt; Sum of mean rates ||
          | |  Peak rate        ---------&gt; f(Peak rates)     ||
          | |  Burst Size       ---------&gt; Sum of Burst sizes||
          | |                    ||    ||                    ||
          | +--------------------+|    |+--------------------+|
          | +--------------------+|    |+--------------------+|
          | |      IP            ||    ||       IP           ||
          | +--------------------+|    |+--------------------+|
          | +--------------------+|    |+--------------------+|
          | | Plaintext Interface||    ||Ciphertext Interface||
          | +--------------------+|    |+--------------------+|
          +-----------------------+    +----------------------+

               Figure 6: Data Flows in a VPN Router Outbound










<span class="grey">Baker &amp; Bose                 Informational                     [Page 24]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-25" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   Parameters on a reservation include:

   Destination Address:  On the plaintext side, the VPN router
      participates in the end-to-end reservations being installed for
      plaintext sessions.  These may include individual flows as
      described in [<a href="./rfc2205" title="&quot;Resource ReSerVation Protocol (RSVP) -- Version 1 Functional Specification&quot;">RFC2205</a>], IPsec data flows [<a href="./rfc2207" title="&quot;RSVP Extensions for IPSEC Data Flows&quot;">RFC2207</a>], aggregate
      reservations [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>], or other types.  It passes an identifier
      for the ciphertext side of the deaggregator to its ciphertext
      unit.

   DSCP:  The DSCP of the plaintext data flow is provided to the cipher
      text side.

   Virtual Port:  The virtual destination port is provided to the cipher
      text side.  This may be derived from an [<a href="./rfc2207" title="&quot;RSVP Extensions for IPSEC Data Flows&quot;">RFC2207</a>] session object
      or from policy information.

   Mean Rate:  The sum of the plaintext mean rates is provided to the
      ciphertext unit.

   Peak Rate:  A function of the plaintext peak rates is provided to the
      ciphertext unit.  This function is less than or equal to the sum
      of the peak rates.

   Burst Size:  The sum of the burst sizes is provided to the cipher
      text unit.

   Messages include:

   Path:  The plaintext PATH message is sent as encrypted data to the
      ciphertext unit.  In parallel, a trigger needs to be sent to the
      ciphertext unit that results in it generating the corresponding
      aggregated PATH message for the ciphertext side.

   Path Error:  This indicates that a PATH message sent to the remote
      enclave was in error.  In the error case, the message itself is
      sent on as encrypted data, but a signal is sent to the ciphertext
      side in case the error affects the ciphertext reservation (such as
      removing or changing state).

   Path Tear:  The PATH Tear message is sent as encrypted data to the
      ciphertext unit.  In parallel, a signal is sent to the cipher text
      side; it will trigger a Path Tear on its reservation in the event
      that this is the last aggregated session, or change the
      SENDER_TSPEC of the aggregated session.






<span class="grey">Baker &amp; Bose                 Informational                     [Page 25]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-26" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   RESV:  The plaintext RESV message is sent as encrypted data to the
      ciphertext unit.  In parallel, a trigger needs to be sent to the
      ciphertext unit that results in it generating the corresponding
      aggregated RESV message for the ciphertext side.

   RESV Error:  This indicates that a RESV message that was received as
      data and forwarded into the enclave was in error or needed to be
      preempted as described in [<a href="./rfc3181" title="&quot;Signaled Preemption Priority Policy Element&quot;">RFC3181</a>] or [<a href="./rfc4495" title="&quot;A Resource Reservation Protocol (RSVP) Extension for the Reduction of Bandwidth of a Reservation Flow&quot;">RFC4495</a>].  In the error
      case, the message itself is sent on as encrypted data, but a
      signal is sent to the ciphertext side in case the error affects
      the ciphertext reservation (such as removing or changing state).

   RESV Tear:  The RESV Tear message is sent as encrypted data to the
      ciphertext unit.  In parallel, a signal is sent to the cipher text
      side; it will trigger a RESV Tear on its reservation in the event
      that this is the last aggregated session, or reduce the bandwidth
      of an existing reservation.

   RESV Confirm:  This indicates that a RESV message received as data
      and forwarded into the enclave, and is now being confirmed.  This
      message is sent as encrypted data to the ciphertext side, and, in
      parallel, a signal is sent to potentially trigger an RESV Confirm
      on the aggregate reservation.




























<span class="grey">Baker &amp; Bose                 Informational                     [Page 26]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-27" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h4"><a class="selflink" id="section-3.1.2" href="#section-3.1.2">3.1.2</a>.  Ciphertext to Plaintext Data Flows</span>

           +-----------------------+    +----------------------+
           | +--------------------+|    |+--------------------+|
           | |RSVP                ||    ||Aggregate RSVP      ||
           | |                    ||    ||  terminated        ||
           | |Per session:        |+    ||                    ||
           | |  Destination       ||    ||                    ||
           | |  Source          &lt;---------Decrypted RSVP      ||
           | |  potential SPI     ||    ||  message sent to   ||
           | |  DSCP              ||    ||  Plaintext unit    ||
           | |  vPort or protocol ||    ||  *as data* for     ||
           | |           and port ||    ||  normal processing ||
           | |  Mean rate         ||    ||                    ||
           | |  Peak rate         ||    ||                    ||
           | |  Burst Size        ||    ||                    ||
           | |                    ||    ||                    ||
           | +--------------------+|    |+--------------------+|
           | +--------------------+|    |+--------------------+|
           | |      IP            ||    ||       IP           ||
           | +--------------------+|    |+--------------------+|
           | +--------------------+|    |+--------------------+|
           | |Plaintext Interface ||    ||Ciphertext Interface||
           | +--------------------+|    |+--------------------+|
           +-----------------------+    +----------------------+

               Figure 7: Data Flows in a VPN Router Inbound

   The aggregate reservation is terminated by the ciphertext side of the
   VPN router.  The RSVP messages related to the subsidiary sessions are
   carried in the encrypted tunnel as data, and therefore arrive at the
   plaintext side with other data.  As the plaintext side participates
   in these reservations, some information is returned to the ciphertext
   size to parameterize the aggregate reservation as described in
   <a href="#section-3.1.1">Section 3.1.1</a> in the processing of the outbound messages.
















<span class="grey">Baker &amp; Bose                 Informational                     [Page 27]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-28" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h3"><a class="selflink" id="section-3.2" href="#section-3.2">3.2</a>.  VPN Routers That Use the Network Guard for Signaling across the</span>
<span class="h3">      Cryptographic Boundary</span>

   As described in <a href="#section-1.6">Section 1.6</a> the Network Guard provides an additional
   path for the reservation signaling between the plaintext and cipher
   text domains.

                                 _.------------.
                            ,--'' Plaintext Domain--.
                         ,-' +--------+  +--------+  `-.
                       ,'    |  Host  |  | Host   |     `.
                     ,'      +--------+  +--------+       `.
                    ;                                       :
                    |         +----------------------+      |
                    :         |  +--------+          |      |
                     `.       |  | Router |          |    ,'
                       `.     |  +---+----+          |  ,'
                         `-   |      +----------+    | ,'
                           ---|    +-+--+  +-+--+--+ |'
                              |----|E/D |--|Net Grd| | VPN Router
                           ,-'|    +-+--+  +-+--+--+ |\
                          ,   |      +----------+    | \
                        ,'    |  +---+----+          |  `.
                      ,'      |  | Router |          |    |
                     /        |  +--------+          |     \
                    ;         +----------------------+      :
                    |                                       |
                    :            Ciphertext Domain          ;

                 Figure 8: RSVP Passage via Network Guard

   In this context, the VPN router is composed of a plaintext router, a
   ciphertext router, an encrypt/decrypt implementation (such as a line
   card or interface device), and a network management process that
   manages the encrypt/decrypt implementation and potentially passes
   defined information flows between the plaintext and ciphertext
   domains.  If the Network Guard is implemented as a software process
   that exchanges configuration instructions between the routers, this
   is simple to understand.  If it is built as a separate systems
   exchanging datagrams, it is somewhat more complex, but conceptually
   equivalent.  For example, the ciphertext router would consider an IP
   datagram received via the Network Guard (control plane) as having
   been received from and concerning the interface used in the data
   plane to the encrypt/decrypt unit.







<span class="grey">Baker &amp; Bose                 Informational                     [Page 28]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-29" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h4"><a class="selflink" id="section-3.2.1" href="#section-3.2.1">3.2.1</a>.  Signaling Flow</span>

   Encrypt/decrypt units may not be capable of terminating and
   originating flows as described in <a href="#section-3.1">Section 3.1</a>, and policy may prevent
   knowledge of the ciphertext network addresses in the plaintext
   router.  In such a case, the plaintext and ciphertext routers may use
   the Network Guard as the path for the signaling flows.  The Network
   Guard performs the following functions to enable the flow of
   reservation signaling across the cryptographic domain

   o  transforms plaintext session identifiers into ciphertext session
      identifiers and vice-versa in IP datagrams and RSVP objects (e.g.
      IP addresses)

   o  performs resource management of aggregated reservations (e.g.,
      including ciphertext encapsulation overhead to resources
      requested)

   o  reads and writes configuration on the encrypt/decrypt units as
      necessary (e.g., reads plaintext to ciphertext IP address mapping)

   In addition, the plaintext and ciphertext routers must support a
   routing function or local interface that ensures that aggregated RSVP
   messages flow via the Network Guard.  However, the signaling flow
   across the entire VPN router at a cryptographic boundary remains
   identical to the description in <a href="#section-3.1">Section 3.1</a>.

   A reader may note that the VPN router described in Figure 8 can be
   collapsed into a single router with two halves, or the Network Guard
   and the encrypt/decrypt units can be part of the plaintext router.
   The details of alternate logical and physical architectures for the
   VPN router are beyond the scope of this document.



















<span class="grey">Baker &amp; Bose                 Informational                     [Page 29]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-30" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


<span class="h4"><a class="selflink" id="section-3.2.2" href="#section-3.2.2">3.2.2</a>.  Use Case with Network Guard</span>

                   ........................................
                   :              VPN Router A            :
                   :                                      :
                   :+-----------++----------++-----------+:
     +------+ RSVP :|           || NetGrd-A ||           |:
     |Host A|&lt;----&gt;:|PT-Router-A|+----------+|CT-Router-A|::::::::
     +------+      :|           ||   E/D-A  ||           |:     ::
                   :+-----------++----------++-----------+:     ::
                   :                A-RSVP                :     ::
                   :            &lt;:::::::::::::&gt;           :     ::
                   :......................................:     ::
                                                         A-RSVP ::
                                                               ,---.
                                                             ,'     `.
                                                            /         \
                                                           ; Interface :
                                                           |  Domain   |
                                                           :           ;
                                                            \         /
                                                             `.     ,'
                                                               '---'
                                                         A-RSVP ::
                   ........................................     ::
                   :              VPN Router B            :     ::
                   :                                      :     ::
                   :+-----------++----------++-----------+:     ::
     +------+ RSVP :|           || NetGrd-B ||           |:     ::
     |Host B|&lt;----&gt;:|PT-Router-B|+----------+|CT-Router-B|::::::::
     +------+      :|           ||   E/D-B  ||           |:
                   :+-----------++----------++-----------+:
                   :                A-RSVP                :
                   :            &lt;:::::::::::::&gt;           :
                   :......................................:

                Figure 9: Aggregated RSVP via Network Guard

   The above figure depicts a simple use case for aggregated signaling
   with the Network Guard.  In this scenario, Host A initiates RSVP
   signaling to Host B for a reservation.  The RSVP signaling between
   the hosts is encapsulated by the VPN routers into encrypted tunnels.
   Aggregated RSVP signaling is triggered by VPN routers, and flows into
   the CT-Routers, as well as the interface domains, to reserve
   resources at RSVP-capable routers on the path.  The aggregation/
   deaggregation point for RSVP reservations in this use case are the
   PT-Routers.  The signaling aggregation of RSVP into A-RSVP at the
   PT-Router is similar to the data flow described in <a href="#section-3.1">Section 3.1</a>.  The



<span class="grey">Baker &amp; Bose                 Informational                     [Page 30]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-31" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   Network Guard performs the additional functions described in <a href="#section-3.2.1">Section</a>
   <a href="#section-3.2.1">3.2.1</a> to transform plaintext A-RSVP messages into suitable ciphertext
   A-RSVP messages.  A typical reservation set up in this case would
   follow these steps.

   o  Host A sends RSVP PATH message to Host B.

   o  PT-Router-A encapsulates RSVP PATH message in encrypted tunnel to
      VPN Router B.

   o  CT Routers and Interface domain carry encrypted RSVP PATH message
      (like any other encrypted data message).

   o  PT-Router-B decrypts RSVP Path Message and sends an E2E PathErr
      message to PT-Router-A in the encrypted tunnel.

   o  PT-Router-B forwards decrypted plaintext RSVP PATH message to Host
      B.

   o  PT-Router-A receives E2E PathErr and sends an aggregated RSVP PATH
      message towards PT-Router-B via the Network Guard.

   o  The NetGrd-A transforms the plaintext aggregate RSVP into the
      ciphertext aggregate RSVP message as described in <a href="#section-3.2.1">Section 3.2.1</a>
      and sends it to the CT-Router-A.

   o  The ciphertext aggregated RSVP message travels through ciphertext
      routers in the interface domain.

   o  CT-Router-B receives the ciphertext aggregate RSVP message and
      sends it to the NetGrd-B.

   o  The NetGrd-B transforms the ciphertext aggregate RSVP into the
      plaintext aggregate RSVP message as described in <a href="#section-3.2.1">Section 3.2.1</a> and
      sends it to the PT-Router-B.

   The subsequent RSVP and Aggregate RSVP signaling follows a similar
   flow, as described in detail in [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] and [<a href="./rfc4860" title="&quot;Generic Aggregate Resource ReSerVation Protocol (RSVP) Reservations&quot;">RFC4860</a>]to aggregate
   each plaintext reservation into a corresponding ciphertext
   reservation.  This ensures that RSVP-capable ciphertext routers
   reserve the required resources for a plaintext end-to-end
   reservation.  Subsequent mechanisms, such as preemption or the
   increase and decrease of resources reserved, may be applied to these
   reservations as described before in this document.  The RSVP data
   flow as described in <a href="#section-3.1">Section 3.1</a> within the VPN router (from the
   plaintext router to the ciphertext router via the Guard) provides
   necessary and sufficient information to routers in the ciphertext
   domain to implement the QoS solution presented in the document.



<span class="grey">Baker &amp; Bose                 Informational                     [Page 31]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-32" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   In this description, we have described the Network Guard as being
   separate from the encrypt/decrypt unit.  This separation exists
   because in certain implementations, it is mandated by those who
   specify the devices.  The separation does not come for free, however;
   the separation of the devices for system-engineering purposes is
   expensive, and it imposes architectural problems.  For example, when
   the Guard is used to aggregate RSVP messages or Protocol Independent
   Multicast (PIM) routing, the traffic is destined to the remote VPN
   router.  This means that the Guard must somehow receive and respond
   to, on behalf of the VPN Router, messages that are not directed to
   it.  Several possible solutions exist; they should be selected
   carefully based on the security and implementation needs of the
   environment.  They are as follows:

   o  In the simplest case, the Network Guard and encrypt/decrypt unit
      can be two independent functions that utilize a common network and
      MAC layer.  This can allow the two functions to share a common MAC
      and IP address, so that traffic destined for one function is also
      received by the other.  In the case that these two functions are
      physically separated on two devices, they can still share a common
      MAC and IP address; however, additional modifications may be
      required on the Guard to filter and not process IP traffic not
      destined for itself.

   o  The ciphertext interface of the Guard could be placed into
      promiscuous mode, allowing it to receive all messages and discard
      all but the few it is interested in.  The security considerations
      on putting a device in promiscuous mode at the VPN boundary needs
      to be taken into account in this method.

   o  The Guard could be engineered to receive all from the ciphertext
      router and pass the bulk of it on to the VPN router through
      another interface.  In this case, the Guard and the VPN router
      would use the same IP address.  This mechanism puts the load of
      all data and management traffic destined for the VPN router upon
      the Guard.

   o  The VPN router could be engineered to receive all traffic from the
      ciphertext router and pass any unencrypted traffic it receives to
      the Guard through another interface.  In this case, the Guard and
      the VPN router would use the same IP address.

   o  All the VPN router functions, as shown in Figure 9, could be
      incorporated into a single chassis, with appropriate internal
      traffic management to send some traffic into the plaintext enclave
      and some to the Guard.  In this case, the Guard and the VPN router
      would be -- at least, functionally -- the same system.




<span class="grey">Baker &amp; Bose                 Informational                     [Page 32]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-33" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   Of these, clearly the last is the simplest architecturally and the
   one that most minimizes the attendant risk.

<span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>.  Security Considerations</span>

   The typical security concerns of datagram integrity, node and user
   authentication are implicitly met by the security association that
   exists between the VPN routers.  The secure data stream that flows
   between the VPN routers is also used for the reservation signaling
   datagrams flowing between VPN routers.  Information that is contained
   in these signaling datagrams receives the same level of encryption
   that is received by the data streams.

   One of the reasons cited for the nesting of VPN routes in <a href="#section-1.3">Section 1.3</a>
   is the different levels of security across the nested VPN routers.
   If the security level decreases from one VPN router to the next VPN
   Router in the nested path, the reservation signaling datagrams will,
   by default, receive the lower security-level treatment.  For most
   cases, the lower security treatment is acceptable.  In certain
   networks, however, the reservation signaling across the entire nested
   path must receive the highest security-level treatment (e.g.,
   encryption, authentication of signaling nodes).  For example, the
   highest precedence level may only be signaled to VPN routers that can
   provide the highest security levels.  If any VPN router in the nested
   path is incapable of providing the highest security level, it cannot
   participate in the reservation mechanism.

   In the general case, the nested path may contain routers that are
   either incapable of participating in VPNs or providing required
   security levels.  These routers can participate in the reservation
   only if the lower security level is acceptable (as configured by
   policy) for the signaling of reservation datagrams.

   VPN routers encapsulate encrypted IP packets and prepend an extra
   header on each packet.  These packets, whether used for signaling or
   data, should be identifiable, at a minimum by the IP addresses and
   DSCP value.  Therefore, the prepended header should contain, at a
   minimum, the DSCP value corresponding to the signaled reservation in
   each packet.  This may literally be the same DSCP as is used for the
   data (forcing control plane traffic to receive the same QoS treatment
   as its data), or a different DSCP that is routed identically
   (separating control and data-plane traffic QoS but not routing).

   Additionally security considerations as described in [<a href="./rfc4860" title="&quot;Generic Aggregate Resource ReSerVation Protocol (RSVP) Reservations&quot;">RFC4860</a>] and
   [<a href="./rfc3175" title="&quot;Aggregation of RSVP for IPv4 and IPv6 Reservations&quot;">RFC3175</a>] are also applicable in this environment; they include the
   integrity of RSVP messages can be ensured via mechanisms described in
   [<a href="./rfc2747" title="&quot;RSVP Cryptographic Authentication&quot;">RFC2747</a>] and [<a href="./rfc3097" title="&quot;RSVP Cryptographic Authentication -- Updated Message Type Value&quot;">RFC3097</a>] and related key management (through manual
   configuration or a key management protocol) at nodes between any



<span class="grey">Baker &amp; Bose                 Informational                     [Page 33]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-34" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   aggregator and deaggregator pair that processes the messages.  In
   addition, confidentiality can be provided between hops by employing
   IPsec.  Further work in the IETF MSEC Working Group may be applicable
   in these environments for key management and confidentiality.

<span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>.  Acknowledgements</span>

   Doug Marquis, James Polk, Mike Tibodeau, Pete Babendreier, Roger
   Levesque, and Subha Dhesikan gave early review comments.

   Comments by Sean O'Keefe, Tony De Simone, Julie Tarr, Chris Christou,
   and their associates resulted in <a href="#section-3.2">Section 3.2</a>.

   Francois Le Faucheur, Bruce Davie, and Chris Christou (with Pratik
   Bose) added [<a href="./rfc4860" title="&quot;Generic Aggregate Resource ReSerVation Protocol (RSVP) Reservations&quot;">RFC4860</a>], which clarified the interaction of this
   approach with the DSCP.

<span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>.  References</span>

<span class="h3"><a class="selflink" id="section-6.1" href="#section-6.1">6.1</a>.  Normative References</span>

   [<a id="ref-RFC2205">RFC2205</a>]       Braden, B., Zhang, L., Berson, S., Herzog, S., and S.
                   Jamin, "Resource ReSerVation Protocol (RSVP) --
                   Version 1 Functional Specification", <a href="./rfc2205">RFC 2205</a>,
                   September 1997.

   [<a id="ref-RFC2207">RFC2207</a>]       Berger, L. and T. O'Malley, "RSVP Extensions for
                   IPSEC Data Flows", <a href="./rfc2207">RFC 2207</a>, September 1997.

   [<a id="ref-RFC2746">RFC2746</a>]       Terzis, A., Krawczyk, J., Wroclawski, J., and L.
                   Zhang, "RSVP Operation Over IP Tunnels", <a href="./rfc2746">RFC 2746</a>,
                   January 2000.

   [<a id="ref-RFC2750">RFC2750</a>]       Herzog, S., "RSVP Extensions for Policy Control", <a href="./rfc2750">RFC</a>
                   <a href="./rfc2750">2750</a>, January 2000.

   [<a id="ref-RFC2996">RFC2996</a>]       Bernet, Y., "Format of the RSVP DCLASS Object", <a href="./rfc2996">RFC</a>
                   <a href="./rfc2996">2996</a>, November 2000.

   [<a id="ref-RFC3175">RFC3175</a>]       Baker, F., Iturralde, C., Le Faucheur, F., and B.
                   Davie, "Aggregation of RSVP for IPv4 and IPv6
                   Reservations", <a href="./rfc3175">RFC 3175</a>, September 2001.

   [<a id="ref-RFC4495">RFC4495</a>]       Polk, J. and S. Dhesikan, "A Resource Reservation
                   Protocol (RSVP) Extension for the Reduction of
                   Bandwidth of a Reservation Flow", <a href="./rfc4495">RFC 4495</a>, May 2006.





<span class="grey">Baker &amp; Bose                 Informational                     [Page 34]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-35" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   [<a id="ref-RFC4542">RFC4542</a>]       Baker, F. and J. Polk, "Implementing an Emergency
                   Telecommunications Service (ETS) for Real-Time
                   Services in the Internet Protocol Suite", <a href="./rfc4542">RFC 4542</a>,
                   May 2006.

   [<a id="ref-RFC4860">RFC4860</a>]       Le Faucheur, F., Davie, B., Bose, P., Christou, C.,
                   and M. Davenport, "Generic Aggregate Resource
                   ReSerVation Protocol (RSVP) Reservations", <a href="./rfc4860">RFC 4860</a>,
                   May 2007.

<span class="h3"><a class="selflink" id="section-6.2" href="#section-6.2">6.2</a>.  Informative References</span>

   [<a id="ref-ITU.MLPP.1990">ITU.MLPP.1990</a>] International Telecommunications Union, "Multilevel
                   Precedence and Preemption Service", ITU-T
                   Recommendation I.255.3, 1990.

   [<a id="ref-RFC0791">RFC0791</a>]       Postel, J., "Internet Protocol", STD 5, <a href="./rfc791">RFC 791</a>,
                   September 1981.

   [<a id="ref-RFC1633">RFC1633</a>]       Braden, B., Clark, D., and S. Shenker, "Integrated
                   Services in the Internet Architecture: an Overview",
                   <a href="./rfc1633">RFC 1633</a>, June 1994.

   [<a id="ref-RFC2209">RFC2209</a>]       Braden, B. and L. Zhang, "Resource ReSerVation
                   Protocol (RSVP) -- Version 1 Message Processing
                   Rules", <a href="./rfc2209">RFC 2209</a>, September 1997.

   [<a id="ref-RFC2210">RFC2210</a>]       Wroclawski, J., "The Use of RSVP with IETF Integrated
                   Services", <a href="./rfc2210">RFC 2210</a>, September 1997.

   [<a id="ref-RFC2460">RFC2460</a>]       Deering, S. and R. Hinden, "Internet Protocol,
                   Version 6 (IPv6) Specification", <a href="./rfc2460">RFC 2460</a>, December
                   1998.

   [<a id="ref-RFC2474">RFC2474</a>]       Nichols, K., Blake, S., Baker, F., and D. Black,
                   "Definition of the Differentiated Services Field (DS
                   Field) in the IPv4 and IPv6 Headers", <a href="./rfc2474">RFC 2474</a>,
                   December 1998.

   [<a id="ref-RFC2475">RFC2475</a>]       Blake, S., Black, D., Carlson, M., Davies, E., Wang,
                   Z., and W. Weiss, "An Architecture for Differentiated
                   Services", <a href="./rfc2475">RFC 2475</a>, December 1998.

   [<a id="ref-RFC2747">RFC2747</a>]       Baker, F., Lindell, B., and M. Talwar, "RSVP
                   Cryptographic Authentication", <a href="./rfc2747">RFC 2747</a>, January
                   2000.





<span class="grey">Baker &amp; Bose                 Informational                     [Page 35]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-36" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


   [<a id="ref-RFC2872">RFC2872</a>]       Bernet, Y. and R. Pabbati, "Application and Sub
                   Application Identity Policy Element for Use with
                   RSVP", <a href="./rfc2872">RFC 2872</a>, June 2000.

   [<a id="ref-RFC3097">RFC3097</a>]       Braden, R. and L. Zhang, "RSVP Cryptographic
                   Authentication -- Updated Message Type Value", <a href="./rfc3097">RFC</a>
                   <a href="./rfc3097">3097</a>, April 2001.

   [<a id="ref-RFC3181">RFC3181</a>]       Herzog, S., "Signaled Preemption Priority Policy
                   Element", <a href="./rfc3181">RFC 3181</a>, October 2001.

   [<a id="ref-RFC3182">RFC3182</a>]       Yadav, S., Yavatkar, R., Pabbati, R., Ford, P.,
                   Moore, T., Herzog, S., and R.  Hess, "Identity
                   Representation for RSVP", <a href="./rfc3182">RFC 3182</a>, October 2001.

   [<a id="ref-RFC3246">RFC3246</a>]       Davie, B., Charny, A., Bennet, J., Benson, K., Le
                   Boudec, J., Courtney, W., Davari, S., Firoiu, V., and
                   D.  Stiliadis, "An Expedited Forwarding PHB (Per-Hop
                   Behavior)", <a href="./rfc3246">RFC 3246</a>, March 2002.

   [<a id="ref-RFC3261">RFC3261</a>]       Rosenberg, J., Schulzrinne, H., Camarillo, G.,
                   Johnston, A., Peterson, J., Sparks, R., Handley, M.,
                   and E.  Schooler, "SIP: Session Initiation Protocol",
                   <a href="./rfc3261">RFC 3261</a>, June 2002.

   [<a id="ref-RFC3312">RFC3312</a>]       Camarillo, G., Marshall, W., and J.  Rosenberg,
                   "Integration of Resource Management and Session
                   Initiation Protocol (SIP)", <a href="./rfc3312">RFC 3312</a>, October 2002.

   [<a id="ref-RFC3473">RFC3473</a>]       Berger, L., "Generalized Multi-Protocol Label
                   Switching (GMPLS) Signaling Resource ReserVation
                   Protocol-Traffic Engineering (RSVP-TE) Extensions",
                   <a href="./rfc3473">RFC 3473</a>, January 2003.

   [<a id="ref-RFC3474">RFC3474</a>]       Lin, Z. and D. Pendarakis, "Documentation of IANA
                   assignments for Generalized MultiProtocol Label
                   Switching (GMPLS) Resource Reservation Protocol -
                   Traffic Engineering (RSVP-TE) Usage and Extensions
                   for Automatically Switched Optical Network (ASON)",
                   <a href="./rfc3474">RFC 3474</a>, March 2003.

   [<a id="ref-RFC4301">RFC4301</a>]       Kent, S. and K. Seo, "Security Architecture for the
                   Internet Protocol", <a href="./rfc4301">RFC 4301</a>, December 2005.

   [<a id="ref-RFC4303">RFC4303</a>]       Kent, S., "IP Encapsulating Security Payload (ESP)",
                   <a href="./rfc4303">RFC 4303</a>, December 2005.





<span class="grey">Baker &amp; Bose                 Informational                     [Page 36]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-37" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


Authors' Addresses

   Fred Baker
   Cisco Systems
   1121 Via Del Rey
   Santa Barbara, California  93117
   USA

   Phone: +1-408-526-4257
   Fax:   +1-413-473-2403
   EMail: fred@cisco.com


   Pratik Bose
   Lockheed Martin
   700 North Frederick Ave
   Gaithersburg, Maryland  20871
   USA

   Phone: +1-301-240-7041
   Fax:   +1-301-240-5748
   EMail: pratik.bose@lmco.com





























<span class="grey">Baker &amp; Bose                 Informational                     [Page 37]</span></pre>
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-38" ></span>
<span class="grey"><a href="./rfc4923">RFC 4923</a>                  QoS in a Nested VPN                August 2007</span>


Full Copyright Statement

   Copyright (C) The IETF Trust (2007).

   This document is subject to the rights, licenses and restrictions
   contained in <a href="https://www.rfc-editor.org/bcp/bcp78">BCP 78</a>, and except as set forth therein, the authors
   retain all their rights.

   This document and the information contained herein are provided on an
   "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
   OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND
   THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS
   OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF
   THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
   WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in <a href="https://www.rfc-editor.org/bcp/bcp78">BCP 78</a> and <a href="https://www.rfc-editor.org/bcp/bcp79">BCP 79</a>.

   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
   <a href="http://www.ietf.org/ipr">http://www.ietf.org/ipr</a>.

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

Acknowledgement

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







Baker &amp; Bose                 Informational                     [Page 38]
</pre>