One document matched: draft-ietf-tsvwg-admitted-realtime-dscp-03.xml


<?xml version="1.0" encoding="US-ASCII"?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?xml-stylesheet type='text/xsl' href='rfc2629.xslt' ?>
<!-- Some of the more generally applicable PIs that most I-Ds might want to use -->
<!-- Try to enforce the ID-nits conventions and DTD validity -->
<?rfc strict="yes" ?>
<!-- Items used when reviewing the document -->
<?rfc comments="no" ?>
<!-- Controls display of  <cref>  elements -->
<?rfc inline="no" ?>
<!-- When no, put comments at end in comments section,
                                 otherwise, put inline -->
<?rfc editing="no" ?>
<!-- When yes, insert editing marks: editing marks consist of a 
                                 string such as  <29>  printed in the blank line at the 
                                 beginning of each paragraph of text. -->
<!-- Create Table of Contents (ToC) and set some options for it.  
         Note the ToC may be omitted for very short documents,but idnits insists on a ToC 
         if the document has more than 15 pages. -->
<?rfc toc="yes"?>
<?rfc tocompact="yes"?>
<!-- If "yes" eliminates blank lines before main section entries. -->
<?rfc tocdepth="3"?>
<!-- Sets the number of levels of sections/subsections... in ToC -->
<!-- Choose the options for the references. 
         Some like symbolic tags in the references (and citations) and others prefer 
         numbers. The RFC Editor always uses symbolic tags.
         The tags used are the anchor attributes of the references. -->
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes" ?>
<!-- If "yes", causes the references to be sorted in order of tags.
                                 This doesn't have any effect unless symrefs is "yes" also. -->
<!-- These two save paper: Just setting compact to "yes" makes savings by not starting each 
         main section on a new page but does not omit the blank lines between list items. 
         If subcompact is also "yes" the blank lines between list items are also omitted. -->
<?rfc compact="yes" ?>
<?rfc subcompact="no" ?>
<!-- end of list of popular I-D processing instructions -->
<!-- end of list of processing instructions -->
<rfc category="bcp" docName="draft-ietf-tsvwg-admitted-realtime-dscp-03"
     ipr="full3978" updates="4542,4594">
  <front>
    <title abbrev="">DSCPs for Capacity-Admitted Traffic</title>

    <author fullname="Fred Baker" initials="F.J." surname="Baker">
      <organization>Cisco Systems</organization>

      <address>
        <postal>
          <street></street>

          <city>Santa Barbara</city>

          <code>93117</code>

          <region>California</region>

          <country>USA</country>
        </postal>

        <phone>+1-408-526-4257</phone>

        <email>fred@cisco.com</email>
      </address>
    </author>

    <author fullname="James Polk" initials="J.M." surname="Polk">
      <organization>Cisco Systems</organization>

      <address>
        <postal>
          <street></street>

          <city>Richardson</city>

          <code>75082</code>

          <region>Texas</region>

          <country>USA</country>
        </postal>

        <phone>+1-817-271-3552</phone>

        <email>jmpolk@cisco.com</email>
      </address>
    </author>

    <author fullname="Martin Dolly" initials="M." surname="Dolly">
      <organization>AT&T Labs</organization>

      <address>
        <postal>
          <street></street>

          <city>Middletown Township</city>

          <code>07748</code>

          <region>New Jersey</region>

          <country>USA</country>
        </postal>

        <phone>+1-732-420-4574</phone>

        <email>mdolly@att.com</email>
      </address>
    </author>

    <date year="2007" />

    <area>Transport Area</area>

    <workgroup>Transport Working Group</workgroup>

    <abstract>
      <t>This document requests one DSCP from the IANA for real-time traffic
      classes similar to voice conforming to the Expedited Forwarding Per Hop
      Behavior, and admitted using a CAC procedure involving authentication,
      authorization, and capacity admission, as compared to a class of
      real-time traffic conforming to the Expedited Forwarding Per Hop
      Behavior but not subject to capacity admission or subject to very coarse
      capacity admission.</t>

      <t>It also recommends that certain classes of video traffic described in
      RFC 4594 and which have similar requirements be require admission using
      a CAC procedure involving authentication, authorization, and capacity
      admission.</t>

      <t>One of the reasons behind this is the need for classes of traffic
      that are handled under special policies, such as the non-preemptive
      Emergency Telecommunication Service, the US DoD's Assured Service (which
      is similar to MLPP), or e-911. Capacity-admitted traffic classes need
      separation from traffic not subject to admission control, from which
      they can deterministically obtain their service requirements, including
      SLA matters.</t>
    </abstract>

    <!--		
		 <note title="Foreword"> 
		 </note> 
		-->

    <note title="Requirements Language">
      <t>The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
      "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
      document are to be interpreted as described in <xref target="RFC2119">
      RFC 2119 </xref>.</t>
    </note>

    <!--
             <?rfc needLines="10" ?> 
             <texttable anchor='table_example' title="A Very Simple Table"> 
                 <preamble> Tables use ttcol to define column headers and widths.
                Every cell then has a "c" element for its content. </preamble> 
  <ttcol align='center'> ttcol #1 </ttcol> 
                     <ttcol align='center'> ttcol #2 </ttcol> 
                       <c> c #1 </c> 		 <c> c #2 </c> 
                       <c> c #3 </c> 		 <c> c #4 </c> 
                       <c> c #5 </c> 		 <c> c #6 </c> 
                 <postamble> which is a very simple example. </postamble> 
             </texttable> 
    -->
  </front>

  <middle>
    <!--		
			 <t> There are multiple list styles: 'symbols', 'letters', 'numbers',
'hanging', 'format', etc. </t> 
			 <t> 
				 <list style="symbols"> 
					 <t> First bullet </t> 
					 <t> Second bullet </t> 
				 </list> 
			 </t> 
-->

    <!--
<figure anchor="xml_happy" title="Figure N"> 
<artwork align="center"> 
<![CDATA[
	ASCII artwork goes here... 
]]> 
</artwork> 
</figure> 
-->

    <section title="Introduction">
      <t>This document requests one DSCP from the IANA for a class of
      real-time traffic conforming to the <xref target="RFC3246"> Expedited
      Forwarding </xref> <xref target="RFC3247"></xref> Per Hop Behavior and
      admitted using a CAC procedure involving authentication, authorization,
      and capacity admission, as compared to a class of real-time traffic
      conforming to the Expedited Forwarding Per Hop Behavior but not subject
      to capacity admission or subject to very coarse capacity admission.</t>

      <t>It also recommends that certain classes of video described in <xref
      target="RFC4594"></xref> be treated as requiring capacity admission as
      well.</t>

      <t>These applications have one or more potential congestion points
      between the video distribution/conferencing bridge or gaming server and
      the user(s), and reserving capacity for them is important to application
      performance. All of these applications have low tolerance to jitter (aka
      delay variation) and loss, as summarized in <xref
      target="implementation"></xref>, and most (except for multimedia
      conferencing) have inelastic flow behavior from Figure 1 of <xref
      target="RFC4594"></xref>. Inelastic flow behavior and low jitter/loss
      tolerance are the service characteristics that define the need for
      admission control behavior.</t>

      <t>One of the reasons behind this is the need for classes of traffic
      that are handled under special policies, such as the non-preemptive
      Emergency Telecommunication Service, the US DoD's Assured Service (which
      is similar to MLPP and uses preemption), or e-911, in addition to normal
      routine calls that use call admission. It is possible to use control
      plane protocols to generally restrict session admission such that
      admitted traffic should receive the desired service, and the policy
      (e.g. Routine, National Security or Emergency Preparedness [NS/EP]
      communications, e-911, etc) need not be signaled in a DSCP. However,
      service providers need to distinguish between special-policy traffic and
      other classes, particularly the existing VoIP services that perform no
      capacity admission or only very coarse capacity admission and can exceed
      their allocated resources.</t>

      <t>The requested DSCP applies to the Telephony Service Class described
      in <xref target="RFC4594"></xref>. The video classes addressed include
      the <list style="symbols">
          <t>interactive real-time traffic (CS4, used for Video conferencing
          and Interactive gaming),</t>

          <t>broadcast TV (CS3) for use in a video on demand context, and</t>

          <t>AF4 Multimedia conferencing (video conferencing).</t>
        </list></t>

      <t>Since the admitted video classes have not had the history of mixing
      admitted and non-admitted traffic in the same PHB as has occurred for
      EF, an additional DSCP code point is not recommended. Instead, the
      recommended "best practice" is to perform admission control for the
      above video classes.</t>

      <t>Other video classes are not believed to be required by the targeted
      services and to not have the current problem of confusion with
      unadmitted traffic. Within an ISP and on inter-ISP links (i.e. within
      networks whose internal paths are uniform at hundreds of megabits or
      faster), one would expect all of this traffic to be carried in the Real
      Time Traffic Class described in <xref
      target="I-D.ietf-tsvwg-diffserv-class-aggr"></xref>.</t>

      <section title="Definitions">
        <t>The following terms and acronyms are used in this document.</t>

        <t><list style="hanging">
            <t hangText="PHB:">A Per-Hop-Behavior (PHB) is the externally
            observable forwarding behavior applied at a DS-compliant node to a
            DS behavior aggregate <xref target="RFC2475"></xref>. It may be
            thought of as a program configured on the interface of an Internet
            host or router, specified drop probabilities, queuing priorities
            or rates, and other handling characteristics for the traffic
            class.</t>

            <t hangText="DSCP:">The Differentiated Services Codepoint (DSCP),
            as defined in <xref target="RFC2474"></xref>, is a value which is
            encoded in the DS field, and which each DS Node MUST use to select
            the PHB which is to be experienced by each packet it forwards
            <xref target="RFC3260"></xref>. It is a 6-bit number embedded into
            the 8-bit TOS field of an IPv4 datagram or the Traffic Class field
            of an IPv6 datagram.</t>

            <t hangText="CAC:">Call Admission Control, which includes concepts
            of authorization (an identified and authenticated user is
            determined to also be authorized to use the service) and capacity
            admission (at the present time, under some stated policy, capacity
            exists to support the call). In the Internet, these are separate
            functions, while in the PSTN they and call routing are carried out
            together.</t>

            <t hangText="UNI:">A User/Network Interface (UNI) is the interface
            (often a physical link or its virtual equivalent) that connects
            two entities that do not trust each other, and in which one (the
            user) purchases connectivity services from the other (the
            network). <xref target="UNI-NNI"></xref> shows two user networks
            connected by what appears to each of them to be a single network
            ("The Internet", access to which is provided by their service
            provider) which provides connectivity services to other users.
            <vspace blankLines="1" /> UNIs tend to be the bottlenecks in the
            Internet, where users purchase relatively low amounts of bandwidth
            for cost or service reasons, and as a result are most subject to
            congestion issues and therefore issues requiring traffic
            conditioning and service prioritization.</t>

            <t hangText="NNI:">A Network/Network Interface (NNI) is the
            interface (often a physical link or its virtual equivalent) that
            connects two entities that trust each other within limits, and in
            which the two are seen as trading services for value. <xref
            target="UNI-NNI"></xref> shows three service networks that
            together provide the connectivity services that we call "the
            Internet". They are different administrations and are very
            probably in competition, but exchange contracts for connectivity
            and capacity that enable them to offer specific services to their
            customers. <vspace blankLines="1" /> NNIs may not be bottlenecks
            in the Internet if service providers contractually agree to
            provision excess capacity at them. However, NNI performance may
            differ by ISP, and the performance guarantee interval may also
            range from a month to a much shorter period of time. Furthermore,
            a peering point NNI may not have contractual performance
            guarantees or may become overloaded during certain conditions.
            They are also policy-controlled interfaces, especially in BGP. As
            a result, they may require policy to be applied in traffic
            prioritization.</t>

            <t hangText="Queue:">There are multiple ways to build a
            multi-queue scheduler. Weighted Round Robin (WRR) literally builds
            multiple lists and visits them in a specified order, while a
            calendar queue (often used to implement Weighted Fair Queuing, or
            WFQ) builds a list for each time interval and enqueues at most a
            stated amount of data in each such list for transmission during
            that time interval. While these differ dramatically in
            implementation, the external difference in behavior is generally
            negligible when they are properly configured. Consistent with the
            definitions used in the <xref target="RFC2475"> Differentiated
            Services Architecture </xref>, these are treated as equivalent in
            this document, and the lists of WRR and the classes of a calendar
            queue will be referred to uniformly as "queues".</t>
          </list></t>

        <figure anchor="UNI-NNI" title="UNI and NNI interfaces">
          <artwork align="center"><![CDATA[  
                                 _.--------.
                             ,-''           `--.
                          ,-'                   `-.
    ,-------.           ,',-------.                `.
  ,'         `.       ,','         `.                `.
 /  User       \ UNI / /   Service   \                 \
(    Network    +-----+    Network    )                 `.
 \             /  ;    \             /                    :
  `.         ,'   ;     `.         .+                     :
    '-------'    /        '-------'  \ NNI                 \
                ;                     \                     :
                ;     "The Internet"   \  ,-------.         :
               ;                        +'         `.        :
 UNI: User/Network Interface           /   Service   \       |
              |                       (    Network    )      |
 NNI: Network/Network Interface        \             /       |
               :                        +.         ,'        ;
                :                      /  '-------'         ;
                :                     /                     ;
    ,-------.    \        ,-------.  / NNI                 /
  ,'         `.   :     ,'         `+                     ;
 /  User       \ UNI   /   Service   \                    ;
(    Network    +-----+    Network    )                 ,'
 \             /     \ \             /                 /
  `.         ,'       `.`.         ,'                ,'
    '-------'           `.'-------'                ,'
                          `-.                   ,-'
                             `--.           _.-'
                                 `--------''
  ]]></artwork>
        </figure>
      </section>

      <section title="Problem">
        <t>In short, the Telephony Service Class described in <xref
        target="RFC4594"></xref> permits the use of capacity admission in
        implementing the service, but present implementations either provide
        no capacity admission services or do so in a manner that depends on
        specific traffic engineering. In the context of the Internet backbone,
        the two are essentially equivalent; the edge network is depending on
        specific engineering by the service provider that may not be
        present.</t>

        <t>However, services are being requested of the network that would
        specifically make use of capacity admission, and would distinguish
        among users or the uses of available Voice-on-IP or Video-on-IP
        capacity in various ways. Various agencies would like to provide
        services as described in section 2.6 of <xref target="RFC4504"></xref>
        or in <xref target="RFC4190"></xref>. This requires the use of
        capacity admission to differentiate among users (which might be 911
        call centers, other offices with preferential service contracts, or
        individual users gaining access with special credentials) to provide
        services to them that are not afforded to routine customer-to-customer
        IP telephony sessions.</t>
      </section>

      <section title="Proposed Solution">
        <t>The IETF is asked to differentiate, in the Telephony Service,
        between sessions that are originated without capacity admission or
        using traffic engineering and sessions that are originated using more
        robust capacity admission procedures. Sessions of the first type use a
        traffic class in which they compete without network-originated control
        as described in <xref target="CAC-LAN"></xref> or <xref
        target="CAC-H323"></xref>, and in the worst case lose traffic due to
        policing. Sessions of the second type cooperate with network control,
        and may be given different levels of preference depending on the
        policies that the network applies. In order to provide this
        differentiation, the IETF requests that the IANA assign a separate
        DSCP value to admitted sessions using the Telephony service (see <xref
        target="IANA"></xref> ).</t>
      </section>
    </section>

    <section anchor="implementation"
             title="Implementation of the Admitted Service Classes">
      <section anchor="data-plane"
               title="Potential implementations of EF in this model">
        <t>There are at least two possible ways to implement the Expedited
        Forwarding PHB in this model. They are to implement separate classes
        as a set of <list style="symbols">
            <t>Multiple data plane traffic classes, each consisting of a
            policer and a queue, and the queues enjoying different priorities,
            or</t>

            <t>Multiple data plane traffic classes, each consisting of a
            policer but feeding into a common queue or multiple queues at the
            same priority.</t>
          </list> We will explain the difference, and describe in what way
        they differ in operation. The reason this is necessary is that there
        is current confusion in the industry, including a widely reported test
        for NS/EP services that implemented the policing model and described
        it as an implementation of the multi-priority model, and discussion in
        other environments of the intermixing of voice and video traffic at
        relatively low bandwidths in the policing model.</t>

        <t>The multi-priority model is shown in <xref
        target="priority-schematic"></xref>. In this model, traffic from each
        service class is placed into a separate priority queue. If data is
        present in both queues, traffic from one of them will always be
        selected for transmission. This has the effect of transferring jitter
        from the higher priority queue to the lower priority queue, and
        reordering traffic in a way that gives the higher priority traffic a
        smaller average queuing delay. Each queue must have its own policer,
        however, to protect the network from errors and attacks; if a traffic
        class thinks it is carrying a certain data rate but an abuse sends
        significantly more, the effect of simple prioritization would not
        preserve the lower priorities of traffic, which could cause routing to
        fail or otherwise impact an SLA.</t>

        <figure anchor="priority-schematic"
                title="Implementation as a data plane priority">
          <artwork align="center"><![CDATA[  
                                  .
          policers    priorities  |`.
EF ---------> <=> ----------||----+  `.
                              high|    `.
EF2---------> <=> ----------||----+     .'-----------
                .             medium  .'
   rate queues  |`.         +-----+ .' Priority
AF1------>||----+  `.      /  low |'   Scheduler
                |    `.   /
AF2------>||----+     .'-+
                |   .'
CS0------>||----+ .' Rate Scheduler
                |'   (WFQ, WRR, etc)
  ]]></artwork>
        </figure>

        <t>The multi-policer model is shown in <xref
        target="policer-schematic"></xref>. In this model, traffic from each
        service class is policed according to its SLA requirements, and then
        placed into a common priority queue. Unlike the multi-priority model,
        the jitter experienced by the traffic classes in this case is the
        same, as there is only one queue, but the sum of the traffic in this
        higher priority queue experiences less average jitter than the elastic
        traffic in the lower priority.</t>

        <figure anchor="policer-schematic"
                title="Implementation as a data plane policer">
          <artwork align="center"><![CDATA[  
          policers    priorities  .
EF ---------> <=> -------\        |`.
                          --||----+  `.
EF2---------> <=> -------/    high|    `.
                .                 |     .'--------
   rate queues  |`.         +-----+   .'
AF1------>||----+  `.      /  low | .' Priority
                |    `.   /       |'   Scheduler
AF2------>||----+     .'-+
                |   .'
CS0------>||----+ .' Rate Scheduler
                |'   (WFQ, WRR, etc)
  ]]></artwork>
        </figure>

        <t>The difference between the two operationally is, as stated, the
        issues of loss due to policing and distribution of jitter.</t>

        <t>If the two traffic classes are, for example, voice and video,
        datagrams containing video data are relatively large (generally the
        size of the path MTU) while datagrams containing voice are relatively
        small, on the order of only 40 to 200 bytes, depending on the codec.
        On lower speed links (less than 10 MBPS), the jitter introduced by
        video to voice can be disruptive, while at higher speeds the jitter is
        nominal compared to the jitter requirements of voice. At access
        network speeds, therefore, <xref target="RFC4594"></xref> recommends
        separation of video and voice into separate queues, while at optical
        speeds <xref target="I-D.ietf-tsvwg-diffserv-class-aggr"></xref>
        recommends that they use a common queue.</t>

        <t>If, on the other hand, the two traffic classes are carrying the
        same type of application with the same jitter requirements, then
        giving one preference in this sense does not benefit the higher
        priority traffic and may harm the lower priority traffic. In such a
        case, using separate policers and a common queue is a superior
        approach.</t>
      </section>

      <section anchor="CAC" title="Capacity admission control">
        <t>There are five major ways that capacity admission is done or has
        been proposed to be done in real-time applications: <list
            style="symbols">
            <t>Capacity admission control by assumption,</t>

            <t>Capacity admission control by call counting,</t>

            <t>End-point capacity admission performed by probing the
            network,</t>

            <t>Centralized capacity admission control, and</t>

            <t>Distributed capacity admission control</t>
          </list></t>

        <section anchor="CAC-LAN"
                 title="Capacity admission control by assumption">
          <t>The first approach is to ignore the matter entirely. If one
          assumes that the capacity available to the application is uniformly
          far in excess of its requirements, it is perhaps overhead that can
          be ignored. This assumption is currently made in Internet VoIP
          offerings such as Skype and Vonage; the end user is responsible to
          place his service on a LAN connected to the Internet backbone by a
          high speed broadband connection and use capable ISPs to deliver the
          service. There is an authorization step in the sense that the
          service ensures that the user pays his bills, but no capacity
          admission is considered because there is a clear separation from the
          application service provider admitting the calls and the access
          network provider admitting the traffic. The two have no way of
          knowing about each other, except maybe in the abstract sense.</t>
        </section>

        <section anchor="CAC-H323"
                 title="Capacity admission control by call counting">
          <t>The H.323 gatekeeper, originally specified in 1996, operates on
          the model that the considerations of <xref target="CAC-LAN"></xref>
          generally apply, and that it is therefore sufficient to count calls
          in order to ensure that any bottlenecks in the network are never
          overloaded. Which phone is calling which phone is configured
          information into the Gatekeeper, ensuring it doesn't admit too many
          calls across a low speed link. The area of influence of a Gatekeeper
          is called a Zone, and limits how far away a Gatekeeper can influence
          calls. This is because call counting doesn't scale when more than
          one server is admitting flows across the same limited speed links.
          This approach is consistent with the original design of H.323, which
          in 1996 was a mechanism for connecting H.320 media gateways across a
          LAN. VoIP has come a long way since then, however, and the
          engineering trade-offs this approach requires in complex networks
          are unsatisfactory.</t>

          <t>SIP provides the option to go down another path, to admit its
          servers at layer 7, have no awareness of lower layer connectivity,
          resulting in a divorce from infrastructure knowledge - save for
          <xref target="RFC3312"></xref>, which binds the two, but only at the
          endpoints.</t>

          <t>In short, if there is a bottleneck anywhere in the network that
          might be used to connect two gatekeepers, SIP proxies that do not
          implement or do not configure the use of <xref
          target="RFC3312"></xref>, or other call management systems, the
          amount of traffic between the two must be contained below that
          bottleneck even if the normal path is of much higher bandwidth. In
          addition, the multiplexing of traffic streams between different
          pairs of gatekeepers over a common LAN infrastructure is not handled
          by the application, and so must be managed in the engineering of the
          network.</t>
        </section>

        <section anchor="CAC-probing"
                 title="End-point capacity admission performed by probing the network">
          <t>The IETF started looking into the use of Pre-Congestion
          Notification mechanism to full fill the need of admission control
          for real-time traffic. The main contribution of this work for
          admission control is to allow the network to provide the network's
          pre-congestion information using encoding of a field in the IP
          header. This network pre-congestion information is then used for
          making admission control decisions. With the decision influenced by
          this network pre-congestion notification information and any
          applicable policy information with possible user credentials and
          situational information. The pre-congestion notification mechanism
          does not limit the placement of the admission control decision point
          or the signaling protocol used.</t>

          <t>The overview of one of the current proposals is provided by <xref
          target="I-D.chan-pcn-problem-statement"></xref>. With the
          pre-congestion notification encoding described in <xref
          target="I-D.briscoe-tsvwg-cl-phb"></xref>. An initial deployment
          model provided by <xref
          target="I-D.briscoe-tsvwg-cl-architecture"></xref>. Another proposal
          is embodied in <xref target="I-D.charny-pcn-single-marking"></xref>.
          Similar approaches have been proposed in <xref
          target="I-D.morita-tsvwg-pps"></xref> and its related drafts, by
          Ivars and Karlsson in their PBAC work, and many others.</t>

          <t>Current simulation work have shown the pre-congestion
          notification mechanism works well with large number of audio and
          video flows on high speed lines. This work is ongoing.</t>
        </section>

        <section anchor="CAC-BB"
                 title="Centralized capacity admission control">
          <t>The concept of a Bandwidth Broker was first discussed in the
          research world surrounding ESNET and Internet II in the late 1990's,
          and has been discussed in the literature pertaining to the <xref
          target="RFC2475"> Differentiated Services Architecture </xref>. It
          is, in short, a central system that performs a variety of services
          on behalf of clients of a network including applying AAA services
          (as in <xref target="RFC2904"></xref> )and authorizing them to use
          specified capacity at specified times. Its strength is that it is
          relatively simple, at least in concept, and can keep track of simple
          book-keeping functions apart from network elements such as routers.
          Its weakness is that it has no idea what the specific routing of any
          stated data flow is, or its capacity apart from services such as
          MPLS Traffic Engineering or engineering assumptions specified by the
          designers of a network, and obtaining that information from the
          network via SNMP GET or other network management action can impose a
          severe network overhead, and is obviously not in real-time.</t>

          <t>For scaling reasons, operational Bandwidth Brokers generally take
          on a semi-distributed or fully distributed nature. They are
          implemented on a per-point-of-presence basis, and in satellite
          networks might be implemented in each terminal. At this point, they
          become difficult to operationally distinguish from distributed
          capacity admission services such as described in <xref
          target="CAC-RSVP"></xref>.</t>
        </section>

        <section anchor="CAC-RSVP"
                 title="Distributed capacity admission control">
          <t>The IETF developed the <xref target="RFC1633"> Integrated
          Services Model </xref> and the <xref target="RFC2205"> RSVP capacity
          admission protocol </xref> in the early 1990's, and then integrated
          it with the Differentiated Services Architecture in <xref
          target="RFC2998"></xref>. Since then, the IETF has been working on a
          next generation signaling protocol called NSIS that can be used for
          capacity admission protocol, and which is limited in scope to
          considering unicast sessions. <xref target="RFC4542"></xref> looked
          at the issue of providing preferential services in the Internet, and
          determined that RSVP with its defined extensions could provide those
          services to unicast and multicast sessions.</t>

          <t>As with the Bandwidth Broker model, there are concerns regarding
          scaling, mentioned in <xref target="RFC2208"></xref>. Present
          implementations that have been measured have been found to not
          display the scaling concerns, however, and in any event the use of
          RSVP Aggregation enables the backbone to handle such sessions in a
          manner similar to an ATM Virtual Path, bundling sessions together
          for capacity management purposes.</t>
        </section>
      </section>

      <section anchor="CAC-priority"
               title="Prioritized capacity admission control">
        <t>Emergency Telecommunication Service, the US DoD's Assured Service,
        and e-911 each call for some form of prioritization of some calls over
        others. Prioritization of the use of bandwidth is fundamentally a
        matter of choices - at a point where one has multiple choices,
        applying a policy that selects among them. In the PSTN, GETS operates
        in favor of an authorized caller either by routing a call that would
        otherwise be refused by a path unavailable to the general public or by
        queuing the call until some existing call completes and bandwidth
        becomes available. e-911 is similar, but the policy is based on the
        called party, the emergency call center. MLPP operates by preempting
        an existing call to make way for the new one.</t>

        <t>In the Internet, routing is not performed on a per-call basis, so,
        apart from interconnections to the PSTN, re-routing isn't an option.
        On the other hand, in the Internet there are more classes of traffic
        than in the PSTN. In the PSTN, all calls are uses of circuits, while
        in the Internet some bandwidth is always reserved for elastic
        applications - at least, it must be available for routing, and there
        is generally significant consideration of the web, instant messaging,
        and other applications. In essence, any capacity admission policy that
        differentiates between calls has the option of temporarily borrowing
        bandwidth from the capacity reserved for elastic traffic by accepting
        new sessions under some prioritized policy while refusing sessions of
        lower priority because the threshold at that priority has been
        reached.</t>

        <t>For example, regardless of the type of capacity admission that is
        used (apart from "no admission process"), one might admit prioritized
        sessions using a higher bandwidth threshold than one admits lower
        priority sessions.</t>

        <t>If capacity admission as described in <xref
        target="CAC-H323"></xref> is in use, the thresholds must be set low
        enough that bandwidth would be available anywhere in the network. This
        greatly limits the utility of such a service due to the level of
        bandwidth waste that results.</t>

        <t>If capacity admission as described in <xref
        target="CAC-probing"></xref> is in use, then either multiple
        thresholds must be applied in marking the traffic, multiple traffic
        marks must be applied, or there must be multiple ways to interpret the
        result. In any event, this is only applicable in domains in which the
        law of large numbers applies.</t>

        <t>If capacity admission as described in <xref target="CAC-BB"></xref>
        is in use, thresholds can be applied related to a general policy or
        SLA, or related to the network ingress and egress in use. It requires
        them to maintain state regarding network traffic routing separate from
        the network; to the extent that is variable, it requires direct
        monitoring in the OSS.</t>

        <t>If capacity admission as described in <xref
        target="CAC-RSVP"></xref> is in use, thresholds can be applied to the
        critical points of the path that the traffic in question actually
        takes because one is asking the equipment that the path traverses.</t>
      </section>
    </section>

    <section title="Recommendations on implementation of an Admitted Telephony Service Class">
      <t>It is the belief of the authors that either data plane PHB described
      in <xref target="data-plane"></xref>, if coupled with adequate AAA and
      capacity admission procedures as described in <xref
      target="CAC-RSVP"></xref>, are sufficient to provide the services
      required for an Admitted Telephony service class and an Admitted
      Multimedia Conferencing Service Class. If preemption is in view, as
      described in section 2.3.5.2 of <xref target="RFC4542"></xref>, this
      provides the tools for carrying out the preemption. If preemption is not
      in view, or in addition to preemptive services, the application of
      different thresholds depending on call precedence has the effect of
      improving the probability of call completion by admitting preferred
      calls at a time that other calls are being refused. Routine and priority
      traffic can be admitted using the same DSCP value, as the choice of
      which calls are admitted is handled in the admission procedure executed
      in the control plane, not the policing of the data plane.</t>

      <t>On the point of what protocols and procedures are required for
      authentication, authorization, and capacity admission, we note that
      clear standards do not at this time exist for bandwidth brokers, NSIS
      has not at this time been finalized and in any event is limited to
      unicast sessions, and that RSVP has been standardized and has the
      relevant services. We therefore recommend the use of RSVP at the UNI.
      Procedures at the NNI are business matters to be discussed between the
      relevant networks, and are recommended but not required.</t>
    </section>

    <section anchor="IANA" title="IANA Considerations">
      <t>This note requests that IANA assign a DSCP value to a second EF
      traffic class consistent with <xref target="RFC3246"></xref> and <xref
      target="RFC3247"></xref>. It implements the Telephony Service Class
      described in <xref target="RFC4594"></xref> at lower speeds and is
      included in the <xref target="I-D.ietf-tsvwg-diffserv-class-aggr">Real
      Time Treatment Aggregate</xref> at higher speeds. The recommended value
      for the code point 101101, paralleling the EF code point, which is
      101110, and is allocated from the pool set aside for experimental use
      but potentially available for standards use as well in <xref
      target="RFC2474"></xref>. The code point should be referred to as
      VOICE-ADMIT.</t>

      <t>This traffic class requires the use of capacity admission such as
      RSVP services together with AAA services at the User/Network Interface
      (UNI); the use of such services at the NNI is at the option of the
      interconnected networks.</t>
    </section>

    <section anchor="Security" title="Security Considerations">
      <t>A major requirement of this service is effective use of a signaling
      protocol such as RSVP, with the capabilities to identify its user either
      as an individual or as a member of some corporate entity, and assert a
      policy such as "routine" or "priority".</t>

      <t>This capability, one has to believe, will be abused by script kiddies
      and others if the proof of identity is not adequately strong or if
      policies are written or implemented improperly by the carriers. This
      goes without saying, but this section is here for it to be said...</t>
    </section>

    <section anchor="Acknowledgements" title="Acknowledgements">
      <t>Kwok Ho Chan offered some textual comments and rewrote <xref
      target="CAC-probing"></xref>. Georgios Karagiannis offered additional
      comments on the same section. The impetus for including Video in the
      discussion, which initially only targeted voice, is from Dave McDysan,
      and text he suggested was included.</t>
    </section>
  </middle>

  <back>
    <!-- references split to informative and normative -->

    <references title="Normative References">
      <?rfc include='reference.RFC.2119'?>

      <?rfc include='reference.RFC.2474'?>

      <?rfc include='reference.RFC.3246'?>
    </references>

    <references title="Informative References">
      <?rfc include='reference.I-D.briscoe-tsvwg-cl-architecture'?>

      <?rfc include='reference.RFC.3247'?>

      <?rfc include='reference.RFC.2475'?>

      <?rfc include='reference.RFC.4542'?>

      <?rfc include='reference.I-D.ietf-tsvwg-diffserv-class-aggr'?>

      <?rfc include='reference.RFC.4594'?>

      <?rfc include='reference.I-D.briscoe-tsvwg-cl-phb'?>

      <?rfc include='reference.RFC.3260'?>

      <?rfc include='reference.I-D.chan-pcn-problem-statement'?>

      <?rfc include='reference.RFC.2998'?>

      <?rfc include='reference.I-D.charny-pcn-single-marking'?>

      <?rfc include='reference.I-D.morita-tsvwg-pps'?>

      <?rfc include='reference.RFC.1633'?>

      <?rfc include='reference.RFC.2205'?>

      <?rfc include='reference.RFC.2208'?>

      <?rfc include='reference.RFC.2904'?>

      <?rfc include='reference.RFC.3312'?>

      <?rfc include='reference.RFC.4190'?>

      <?rfc include='reference.RFC.4504'?>
    </references>
  </back>
</rfc>

PAFTECH AB 2003-20262026-04-23 11:05:25