Network Working Group                                         K. Nichols
Request for Comments: 3086                                 Packet Design
Category: Informational                                     B. Carpenter
                                                                    IBM
                                                             April 2001


      Definition of Differentiated Services Per Domain Behaviors
                  and Rules for their Specification

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 Internet Society (2001).  All Rights Reserved.

Abstract

  The differentiated services framework enables quality-of-service
  provisioning within a network domain by applying rules at the edges
  to create traffic aggregates and coupling each of these with a
  specific forwarding path treatment in the domain through use of a
  codepoint in the IP header.  The diffserv WG has defined the general
  architecture for differentiated services and has focused on the
  forwarding path behavior required in routers, known as "per-hop
  forwarding behaviors" (or PHBs).  The WG has also discussed
  functionality required at diffserv (DS) domain edges to select
  (classifiers) and condition (e.g., policing and shaping) traffic
  according to the rules.  Short-term changes in the QoS goals for a DS
  domain are implemented by changing only the configuration of these
  edge behaviors without necessarily reconfiguring the behavior of
  interior network nodes.

  The next step is to formulate examples of how forwarding path
  components (PHBs, classifiers, and traffic conditioners) can be used
  to compose traffic aggregates whose packets experience specific
  forwarding characteristics as they transit a differentiated services
  domain.  The WG has decided to use the term per-domain behavior, or
  PDB, to describe the behavior experienced by a particular set of
  packets as they cross a DS domain.  A PDB is characterized by
  specific metrics that quantify the treatment a set of packets with a
  particular DSCP (or set of DSCPs) will receive as it crosses a DS
  domain.  A PDB specifies a forwarding path treatment for a traffic
  aggregate and, due to the role that particular choices of edge and



Nichols & Carpenter          Informational                      [Page 1]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  PHB configuration play in its resulting attributes, it is where the
  forwarding path and the control plane interact.  The measurable
  parameters of a PDB should be suitable for use in Service Level
  Specifications at the network edge.

  This document defines and discusses Per-Domain Behaviors in detail
  and lays out the format and required content for contributions to the
  Diffserv WG on PDBs and the procedure that will be applied for
  individual PDB specifications to advance as WG products.  This format
  is specified to expedite working group review of PDB submissions.

Table of Contents

   1. Introduction ................................................ 2
   2. Definitions ................................................. 4
   3. The Value of Defining Edge-to-Edge Behavior ................. 5
   4. Understanding PDBs .......................................... 7
   5. Format for Specification of Diffserv Per-Domain Behaviors ...13
   6. On PDB Attributes ...........................................16
   7. A Reference Per-Domain Behavior .............................19
   8. Guidelines for Advancing PDB Specifications .................21
   9. Security Considerations .....................................22
  10. Acknowledgements ............................................22
      References ..................................................22
      Authors' Addresses ..........................................23
      Full Copyright Statement ....................................24

1 Introduction

  Differentiated Services allows an approach to IP Quality of Service
  that is modular, incrementally deployable, and scalable while
  introducing minimal per-node complexity [RFC2475].  From the end
  user's point of view, QoS should be supported end-to-end between any
  pair of hosts.  However, this goal is not immediately attainable.  It
  will require interdomain QoS support, and many untaken steps remain
  on the road to achieving this.  One essential step, the evolution of
  the business models for interdomain QoS, will necessarily develop
  outside of the IETF.  A goal of the diffserv WG is to provide the
  firm technical foundation that allows these business models to
  develop.  The first major step will be to support edge-to-edge or
  intradomain QoS between the ingress and egress of a single network,
  i.e., a DS Domain in the terminology of RFC 2474.  The intention is
  that this edge-to-edge QoS should be composable, in a purely
  technical sense, to a quantifiable QoS across a DS Region composed of
  multiple DS domains.






Nichols & Carpenter          Informational                      [Page 2]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  The Diffserv WG has finished the first phase of standardizing the
  behaviors required in the forwarding path of all network nodes, the
  per-hop forwarding behaviors or PHBs.  The PHBs defined in RFCs 2474,
  2597 and 2598 give a rich toolbox for differential packet handling by
  individual boxes.  The general architectural model for diffserv has
  been documented in RFC 2475.  An informal router model [MODEL]
  describes a model of traffic conditioning and other forwarding
  behaviors.  However, technical issues remain in moving "beyond the
  box" to intradomain QoS models.

  The ultimate goal of creating scalable end-to-end QoS in the Internet
  requires that we can identify and quantify behavior for a group of
  packets that is preserved when they are aggregated with other packets
  as they traverse the Internet.  The step of specifying forwarding
  path attributes on a per-domain basis for a set of packets
  distinguished only by the mark in the DS field of individual packets
  is critical in the evolution of Diffserv QoS and should provide the
  technical input that will aid in the construction of business models.
  This document defines and specifies the term "Per-Domain Behavior" or
  PDB to describe QoS attributes across a DS domain.

  Diffserv classification and traffic conditioning are applied to
  packets arriving at the boundary of a DS domain to impose
  restrictions on the composition of the resultant traffic aggregates,
  as distinguished by the DSCP marking , inside the domain.  The
  classifiers and traffic conditioners are set to reflect the policy
  and traffic goals for that domain and may be specified in a TCA
  (Traffic Conditioning Agreement).  Once packets have crossed the DS
  boundary, adherence to diffserv principles makes it possible to group
  packets solely according to the behavior they receive at each hop (as
  selected by the DSCP).  This approach has well-known scaling
  advantages, both in the forwarding path and in the control plane.
  Less well recognized is that these scaling properties only result if
  the per-hop behavior definition gives rise to a particular type of
  invariance under aggregation.  Since the per-hop behavior must be
  equivalent for every node in the domain, while the set of packets
  marked for that PHB may be different at every node, PHBs should be
  defined such that their characteristics do not depend on the traffic
  volume of the associated BA on a router's ingress link nor on a
  particular path through the DS domain taken by the packets.
  Specifically, different streams of traffic that belong to the same
  traffic aggregate merge and split as they traverse the network.  If
  the properties of a PDB using a particular PHB hold regardless of how
  the temporal characteristics of the marked traffic aggregate change
  as it traverses the domain, then that PDB scales.  (Clearly this
  assumes that numerical parameters such as bandwidth allocated to the
  particular PDB may be different at different points in the network,
  and may be adjusted dynamically as traffic volume varies.)  If there



Nichols & Carpenter          Informational                      [Page 3]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  are limits to where the properties hold, that translates to a limit
  on the size or topology of a DS domain that can use that PDB.
  Although useful single-link DS domains might exist, PDBs that are
  invariant with network size or that have simple relationships with
  network size and whose properties can be recovered by reapplying
  rules (that is, forming another diffserv boundary or edge to re-
  enforce the rules for the traffic aggregate) are needed for building
  scalable end-to-end quality of service.

  There is a clear distinction between the definition of a Per-Domain
  Behavior in a DS domain and a service that might be specified in a
  Service Level Agreement.  The PDB definition is a technical building
  block that permits the coupling of classifiers, traffic conditioners,
  specific PHBs, and particular configurations with a resulting set of
  specific observable attributes which may be characterized in a
  variety of ways.  These definitions are intended to be useful tools
  in configuring DS domains, but the PDB (or PDBs) used by a provider
  is not expected to be visible to customers any more than the specific
  PHBs employed in the provider's network would be.  Network providers
  are expected to select their own measures to make customer-visible in
  contracts and these may be stated quite differently from the
  technical attributes specified in a PDB definition, though the
  configuration of a PDB might be taken from a Service Level
  Specification (SLS).  Similarly, specific PDBs are intended as tools
  for ISPs to construct differentiated services offerings; each may
  choose different sets of tools, or even develop their own, in order
  to achieve particular externally observable metrics.  Nevertheless,
  the measurable parameters of a PDB are expected to be among the
  parameters cited directly or indirectly in the Service Level
  Specification component of a corresponding SLA.

  This document defines Differentiated Services Per-Domain Behaviors
  and specifies the format that must be used for submissions of
  particular PDBs to the Diffserv WG.

2 Definitions

  The following definitions are stated in RFCs 2474 and 2475 and are
  repeated here for easy reference:

  " Behavior Aggregate: a collection of packets with the same codepoint
    crossing a link in a particular direction.

  " Differentiated Services Domain: a contiguous portion of the
    Internet over which a consistent set of differentiated services
    policies are administered in a coordinated fashion.  A
    differentiated services domain can represent different




Nichols & Carpenter          Informational                      [Page 4]

RFC 3086             Diffserv per Domain Behaviors            April 2001


    administrative domains or autonomous systems, different trust
    regions, different network technologies (e.g., cell/frame), hosts
    and routers, etc.  Also DS domain.

  " Differentiated Services Boundary: the edge of a DS domain, where
    classifiers and traffic conditioners are likely to be deployed.  A
    differentiated services boundary can be further sub-divided into
    ingress and egress nodes, where the ingress/egress nodes are the
    downstream/upstream nodes of a boundary link in a given traffic
    direction.  A differentiated services boundary typically is found
    at the ingress to the first-hop differentiated services-compliant
    router (or network node) that a host's packets traverse, or at the
    egress of the last-hop differentiated services-compliant router or
    network node that packets traverse before arriving at a host.  This
    is sometimes referred to as the boundary at a leaf router.  A
    differentiated services boundary may be co-located with a host,
    subject to local policy.  Also DS boundary.

  To these we add:

  " Traffic Aggregate: a collection of packets with a codepoint that
    maps to the same PHB, usually in a DS domain or some subset of a DS
    domain.  A traffic aggregate marked for the foo PHB is referred to
    as the "foo traffic aggregate" or "foo aggregate" interchangeably.
    This generalizes the concept of Behavior Aggregate from a link to a
    network.

  " Per-Domain Behavior: the expected treatment that an identifiable or
    target group of packets will receive from "edge-to-edge" of a DS
    domain.  (Also PDB.)  A particular PHB (or, if applicable, list of
    PHBs) and traffic conditioning requirements are associated with
    each PDB.

  " A Service Level Specification (SLS) is a set of parameters and
    their values which together define the service offered to a traffic
    stream by a DS domain.  It is expected to include specific values
    or bounds for PDB parameters.

3 The Value of Defining Edge-to-Edge Behavior

  As defined in section 2, a PDB describes the edge-to-edge behavior
  across a DS domain's "cloud." Specification of the transit
  expectations of packets matching a target for a particular diffserv
  behavior across a DS domain will both assist in the deployment of
  single-domain QoS and will help enable the composition of end-to-end,
  cross-domain services.  Networks of DS domains can be connected to
  create end-to-end services by building on the PDB characteristics
  without regard to the particular PHBs used.  This level of



Nichols & Carpenter          Informational                      [Page 5]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  abstraction makes it easier to compose cross-domain services as well
  as making it possible to hide details of a network's internals while
  exposing information sufficient to enable QoS.

  Today's Internet is composed of multiple independently administered
  domains or Autonomous Systems (ASs), represented by the "clouds" in
  figure 1.  To deploy ubiquitous end-to-end quality of service in the
  Internet, business models must evolve that include issues of charging
  and reporting that are not in scope for the IETF.  In the meantime,
  there are many possible uses of quality of service within an AS and
  the IETF can address the technical issues in creating an intradomain
  QoS within a Differentiated Services framework.  In fact, this
  approach is quite amenable to incremental deployment strategies.

  Where DS domains are independently administered, the evolution of the
  necessary business agreements and future signaling arrangements will
  take some time, thus, early deployments will be within a single
  administrative domain.  Putting aside the business issues, the same
  technical issues that arise in interconnecting DS domains with
  homogeneous administration will arise in interconnecting the
  autonomous systems (ASs) of the Internet.

                ----------------------------------------
                |                AS2                   |
                |                                      |
   -------      |     ------------     ------------    |
   | AS1 |------|-----X           |    |          |    |
   -------      |     |           |    Y          |    |        -------
                |     |           |   /|          X----|--------| AS3 |
                |     |           |  / |          |    |        -------
                |     |           | /  ------------    |
                |     |           Y      |             |
                |     |           | \  ------------    |
   -------      |     |           |  \ |          |    |
   | AS4 |------|-----X           |   \|          |    |
   -------      |     |           |    Y          X----|------
                |     |           |    |          |    |
                |     ------------     ------------    |
                |                                      |
                |                                      |
                ----------------------------------------

        Figure 1: Interconnection of ASs and DS Domains

  A single AS (e.g., AS2 in figure 1) may be composed of subnetworks
  and, as the definition allows, these can be separate DS domains.  An
  AS might have multiple DS domains for a number of reasons, most
  notable being to follow topological and/or technological boundaries



Nichols & Carpenter          Informational                      [Page 6]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  and to separate the allocation of resources.  If we confine ourselves
  to the DS boundaries between these "interior" DS domains, we avoid
  the non-technical problems of setting up a service and can address
  the issues of creating characterizable PDBs.

  The incentive structure for differentiated services is based on
  upstream domains ensuring their traffic conforms to the Traffic
  Conditioning Agreements (TCAs) with downstream domains and downstream
  domains enforcing that TCA, thus metrics associated with PDBs can be
  sensibly computed.  The letters "X" and "Y" in figure 1 represent the
  DS boundary routers containing traffic conditioners that ensure and
  enforce conformance (e.g., shapers and policers).  Although policers
  and shapers are expected at the DS boundaries of ASs (the "X" boxes),
  they might appear anywhere, or nowhere, inside the AS.  Specifically,
  the boxes at the DS boundaries internal to the AS (the "Y" boxes) may
  or may not condition traffic.  Technical guidelines for the placement
  and configuration of DS boundaries should derive from the attributes
  of a particular PDB under aggregation and multiple hops.

  This definition of PDB continues the separation of forwarding path
  and control plane described in RFC 2474.  The forwarding path
  characteristics are addressed by considering how the behavior at
  every hop of a packet's path is affected by the merging and branching
  of traffic aggregates through multiple hops.  Per-hop behaviors in
  nodes are configured infrequently, representing a change in network
  infrastructure.  More frequent quality-of-service changes come from
  employing control plane functions in the configuration of the DS
  boundaries.  A PDB provides a link between the DS domain level at
  which control is exercised to form traffic aggregates with quality-
  of-service goals across the domain and the per-hop and per-link
  treatments packets receive that results in meeting the quality-of-
  service goals.

4 Understanding PDBs

4.1 Defining PDBs

  RFCs 2474 and 2475 define a Differentiated Services Behavior
  Aggregate as "a collection of packets with the same DS codepoint
  crossing a link in a particular direction" and further state that
  packets with the same DSCP get the same per-hop forwarding treatment
  (or PHB) everywhere inside a single DS domain.  Note that even if
  multiple DSCPs map to the same PHB, this must hold for each DSCP
  individually.  In section 2 of this document, we introduced a more
  general definition of a traffic aggregate in the diffserv sense so
  that we might easily refer to the packets which are mapped to the
  same PHB everywhere within a DS domain.  Section 2 also presented a
  short definition of PDBs which we expand upon in this section:



Nichols & Carpenter          Informational                      [Page 7]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  Per-Domain Behavior: the expected treatment that an identifiable or
    target group of packets will receive from "edge to edge" of a DS
    domain.  A particular PHB (or, if applicable, list of PHBs) and
    traffic conditioning requirements are associated with each PDB.

  Each PDB has measurable, quantifiable, attributes that can be used to
  describe what happens to its packets as they enter and cross the DS
  domain.  These derive from the characteristics of the traffic
  aggregate that results from application of classification and traffic
  conditioning during the entry of packets into the DS domain and the
  forwarding treatment (PHB) the packets get inside the domain, but can
  also depend on the entering traffic loads and the domain's topology.
  PDB attributes may be absolute or statistical and they may be
  parameterized by network properties.  For example, a loss attribute
  might be expressed as "no more than 0.1% of packets will be dropped
  when measured over any time period larger than T", a delay attribute
  might be expressed as "50% of delivered packets will see less than a
  delay of d milliseconds, 30% will see a delay less than 2d ms, 20%
  will see a delay of less than 3d ms." A wide range of metrics is
  possible.  In general they will be expressed as bounds or percentiles
  rather than as absolute values.

  A PDB is applied to a target group of packets arriving at the edge of
  the DS domain.  The target group is distinguished from all arriving
  packets by use of packet classifiers [RFC2475] (where the classifier
  may be "null").  The action of the PDB on the target group has two
  parts.  The first part is the the use of traffic conditioning to
  create a traffic aggregate.  During traffic conditioning, conformant
  packets are marked with a DSCP for the PHB associated with the PDB
  (see figure 2).  The second part is the treatment experienced by
  packets from the same traffic aggregate transiting the interior of a
  DS domain, between and inside of DS domain boundaries.  The following
  subsections further discuss these two effects on the target group
  that arrives at the DS domain boundary.

          -----------   ------------   --------------------   foo
arriving _|classifiers|_|target group|_|traffic conditioning|_ traffic
packets   |           | |of packets  | |& marking (for foo) |  aggregate
          -----------   ------------   --------------------

        Figure 2: Relationship of the traffic aggregate associated
                   with a PDB to arriving packets









Nichols & Carpenter          Informational                      [Page 8]

RFC 3086             Diffserv per Domain Behaviors            April 2001


4.1.1 Crossing the DS edge: the effects of traffic conditioning on the
     target group

  This effect is quantified by the relationship of the emerging traffic
  aggregate to the entering target group.  That relationship can depend
  on the arriving traffic pattern as well as the configuration of the
  traffic conditioners.  For example, if the EF PHB [RFC2598] and a
  strict policer of rate R are associated with the foo PDB, then the
  first part of characterizing the foo PDB is to write the relationship
  between the arriving target packets and the departing foo traffic
  aggregate.  In this case, "the rate of the emerging foo traffic
  aggregate is less than or equal to the smaller of R and the arrival
  rate of the target group of packets" and additional temporal
  characteristics of the packets (e.g., burst) may be specified as
  desired.  Thus, there is a "loss rate" on the arriving target group
  that results from sending too much traffic or the traffic with the
  wrong temporal characteristics.  This loss rate should be entirely
  preventable (or controllable) by the upstream sender conforming to
  the traffic conditioning associated with the PDB specification.

  The issue of "who is in control" of the loss (or demotion) rate helps
  to clearly delineate this component of PDB performance from that
  associated with transiting the domain.  The latter is completely
  under control of the operator of the DS domain and the former is used
  to ensure that the entering traffic aggregate conforms to the traffic
  profile to which the operator has provisioned the network.  Further,
  the effects of traffic conditioning on the target group can usually
  be expressed more simply than the effects of transiting the DS domain
  on the traffic aggregate's traffic profile.

  A PDB may also apply traffic conditioning at DS domain egress.  The
  effect of this conditioning on the overall PDB attributes would be
  treated similarly to the ingress characteristics (the authors may
  develop more text on this in the future, but it does not materially
  affect the ideas presented in this document.)

4.1.2 Crossing the DS domain: transit effects

  The second component of PDB performance is the metrics that
  characterize the transit of a packet of the PDB's traffic aggregate
  between any two edges of the DS domain boundary shown in figure 3.
  Note that the DS domain boundary runs through the DS boundary routers
  since the traffic aggregate is generally formed in the boundary
  router before the packets are queued and scheduled for output.  (In
  most cases, this distinction is expected to be important.)






Nichols & Carpenter          Informational                      [Page 9]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  DSCPs should not change in the interior of a DS domain as there is no
  traffic conditioning being applied.  If it is necessary to reapply
  the kind of traffic conditioning that could result in remarking,
  there should be a DS domain boundary at that point, though such an
  "interior" boundary can have "lighter weight" rules in its TCA.
  Thus, when measuring attributes between locations as indicated in
  figure 3, the DSCP at the egress side can be assumed to have held
  throughout the domain.

                              -------------
                              |           |
                         -----X           |
                              |           |
                              |   DS      |
                              |   domain  X----
                              |           |
                         -----X           |
                              |           |
                              -------------

         Figure 3: Range of applicability of attributes of a traffic
                   aggregate associated with a PDB (is between the
                   points marked "X")

  Though a DS domain may be as small as a single node, more complex
  topologies are expected to be the norm, thus the PDB definition must
  hold as its traffic aggregate is split and merged on the interior
  links of a DS domain.  Packet flow in a network is not part of the
  PDB definition; the application of traffic conditioning as packets
  enter the DS domain and the consistent PHB through the DS domain must
  suffice.  A PDB's definition does not have to hold for arbitrary
  topologies of networks, but the limits on the range of applicability
  for a specific PDB must be clearly specified.

  In general, a PDB operates between N ingress points and M egress
  points at the DS domain boundary.  Even in the degenerate case where
  N=M=1, PDB attributes are more complex than the definition of PHB
  attributes since the concatenation of the behavior of intermediate
  nodes affects the former.  A complex case with N and M both greater
  than one involves splits and merges in the traffic path and is non-
  trivial to analyze.  Analytic, simulation, and experimental work will
  all be necessary to understand even the simplest PDBs.

4.2 Constructing PDBs

  A DS domain is configured to meet the network operator's traffic
  engineering goals for the domain independently of the performance
  goals for a particular flow of a traffic aggregate.  Once the



Nichols & Carpenter          Informational                     [Page 10]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  interior routers are configured for the number of distinct traffic
  aggregates that the network will handle, each PDB's allocation at the
  edge comes from meeting the desired performance goals for the PDB's
  traffic aggregate subject to that configuration of packet schedulers
  and bandwidth capacity.  The configuration of traffic conditioners at
  the edge may be altered by provisioning or admission control but the
  decision about which PDB to use and how to apply classification and
  traffic conditioning comes from matching performance to goals.

  For example, consider the DS domain of figure 3.  A PDB with an
  explicit bound on loss must apply traffic conditioning at the
  boundary to ensure that on the average no more packets are admitted
  than can emerge.  Though, queueing internal to the network may result
  in a difference between input and output traffic over some
  timescales, the averaging timescale should not exceed what might be
  expected for reasonably sized buffering inside the network.  Thus if
  bursts are allowed to arrive into the interior of the network, there
  must be enough capacity to ensure that losses don't exceed the bound.
  Note that explicit bounds on the loss level can be particularly
  difficult as the exact way in which packets merge inside the network
  affects the burstiness of the PDB's traffic aggregate and hence,
  loss.

  PHBs give explicit expressions of the treatment a traffic aggregate
  can expect at each hop.  For a PDB, this behavior must apply to
  merging and diverging traffic aggregates, thus characterizing a PDB
  requires understanding what happens to a PHB under aggregation.  That
  is, PHBs recursively applied must result in a known behavior.  As an
  example, since maximum burst sizes grow with the number of microflows
  or traffic aggregate streams merged, a PDB specification must address
  this.  A clear advantage of constructing behaviors that aggregate is
  the ease of concatenating PDBs so that the associated traffic
  aggregate has known attributes that span interior DS domains and,
  eventually, farther.  For example, in figure 1 assume that we have
  configured the foo PDB on the interior DS domains of AS2.  Then
  traffic aggregates associated with the foo PDB in each interior DS
  domain of AS2 can be merged at the shaded interior boundary routers.
  If the same (or fewer) traffic conditioners as applied at the
  entrance to AS2 are applied at these interior boundaries, the
  attributes of the foo PDB should continue to be used to quantify the
  expected behavior.  Explicit expressions of what happens to the
  behavior under aggregation, possibly parameterized by node in-degrees
  or network diameters, are necessary to determine what to do at the
  internal aggregation points.  One approach might be to completely
  reapply the traffic conditioning at these points; another might
  employ some limited rate-based remarking only.





Nichols & Carpenter          Informational                     [Page 11]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  Multiple PDBs may use the same PHB.  The specification of a PDB can
  contain a list of PHBs and their required configuration, all of which
  would result in the same PDB.  In operation, it is expected that a
  single domain will use a single PHB to implement a particular PDB,
  though different domains may select different PHBs.  Recall that in
  the diffserv definition [RFC2474], a single PHB might be selected
  within a domain by a list of DSCPs.  Multiple PDBs might use the same
  PHB in which case the transit performance of traffic aggregates of
  these PDBs will, of necessity, be the same.  Yet, the particular
  characteristics that the PDB designer wishes to claim as attributes
  may vary, so two PDBs that use the same PHB might not be specified
  with the same list of attributes.

  The specification of the transit expectations of PDBs across domains
  both assists in the deployment of QoS within a DS domain and helps
  enable the composition of end-to-end, cross-domain services to
  proceed by making it possible to hide details of a domain's internals
  while exposing characteristics necessary for QoS.

4.3 PDBs using PHB Groups

  The use of PHB groups to construct PDBs can be done in several ways.
  A single PHB member of a PHB group might be used to construct a
  single PDB.  For example, a PDB could be defined using just one of
  the Class Selector Compliant PHBs [RFC2474].  The traffic
  conditioning for that PDB and the required configuration of the
  particular PHB would be defined in such a way that there was no
  dependence or relationship with the manner in which other PHBs of the
  group are used or, indeed, whether they are used in that DS domain.
  In this case, the reasonable approach would be to specify this PDB
  alone in a document which expressly called out the conditions and
  configuration of the Class Selector PHB required.

  A single PDB can be constructed using more than one PHB from the same
  PHB group.  For example, the traffic conditioner described in RFC
  2698 might be used to mark a particular entering traffic aggregate
  for one of the three AF1x PHBs [RFC2597] while the transit
  performance of the resultant PDB is specified, statistically, across
  all the packets marked with one of those PHBs.

  A set of related PDBs might be defined using a PHB group.  In this
  case, the related PDBs should be defined in the same document.  This
  is appropriate when the traffic conditioners that create the traffic
  aggregates associated with each PDB have some relationships and
  interdependencies such that the traffic aggregates for these PDBs
  should be described and characterized together.  The transit
  attributes will depend on the PHB associated with the PDB and will
  not be the same for all PHBs in the group, though there may be some



Nichols & Carpenter          Informational                     [Page 12]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  parameterized interrelationship between the attributes of each of
  these PDBs.  In this case, each PDB should have a clearly separate
  description of its transit attributes (delineated in a separate
  subsection) within the document.  For example, the traffic
  conditioner described in RFC 2698 might be used to mark arriving
  packets for three different AF1x PHBs, each of which is to be treated
  as a separate traffic aggregate in terms of transit properties.  Then
  a single document could be used to define and quantify the
  relationship between the arriving packets and the emerging traffic
  aggregates as they relate to one another.  The transit
  characteristics of packets of each separate AF1x traffic aggregate
  should be described separately within the document.

  Another way in which a PHB group might be used to create one PDB per
  PHB might have decoupled traffic conditioners, but some relationship
  between the PHBs of the group.  For example, a set of PDBs might be
  defined using Class Selector Compliant PHBs [RFC2474] in such a way
  that the traffic conditioners that create the traffic aggregates are
  not related, but the transit performance of each traffic aggregate
  has some parametric relationship to the other.  If it makes sense to
  specify them in the same document, then the author(s) should do so.

4.4 Forwarding path vs. control plane

  A PDB's associated PHB, classifiers, and traffic conditioners are all
  in the packet forwarding path and operate at line rates.  PHBs,
  classifiers, and traffic conditioners are configured in response to
  control plane activity which takes place across a range of time
  scales, but, even at the shortest time scale, control plane actions
  are not expected to happen per-packet.  Classifiers and traffic
  conditioners at the DS domain boundary are used to enforce who gets
  to use the PDB and how the PDB should behave temporally.
  Reconfiguration of PHBs might occur monthly, quarterly, or only when
  the network is upgraded.  Classifiers and traffic conditioners might
  be reconfigured at a few regular intervals during the day or might
  happen in response to signalling decisions thousands of times a day.
  Much of the control plane work is still evolving and is outside the
  charter of the Diffserv WG.  We note that this is quite appropriate
  since the manner in which the configuration is done and the time
  scale at which it is done should not affect the PDB attributes.

5 Format for Specification of Diffserv Per-Domain Behaviors

  PDBs arise from a particular relationship between edge and interior
  (which may be parameterized).  The quantifiable characteristics of a
  PDB must be independent of whether the network edge is configured
  statically or dynamically.  The particular configuration of traffic




Nichols & Carpenter          Informational                     [Page 13]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  conditioners at the DS domain edge is critical to how a PDB performs,
  but the act(s) of configuring the edge is a control plane action
  which can be separated from the specification of the PDB.

  The following sections must be present in any specification of a
  Differentiated Services PDB.  Of necessity, their length and content
  will vary greatly.

5.1 Applicability Statement

  All PDB specs must have an applicability statement that outlines the
  intended use of this PDB and the limits to its use.

5.2 Technical specification

  This section specifies the rules or guidelines to create this PDB,
  each distinguished with "may", "must" and "should." The technical
  specification must list the classification and traffic conditioning
  required (if any) and the PHB (or PHBs) to be used with any
  additional requirements on their configuration beyond that contained
  in RFCs.  Classification can reflect the results of an admission
  control process.  Traffic conditioning may include marking, traffic
  shaping, and policing.  A Service Provisioning Policy might be used
  to describe the technical specification of a particular PDB.

5.3 Attributes

  A PDB's attributes tell how it behaves under ideal conditions if
  configured in a specified manner (where the specification may be
  parameterized).  These might include drop rate, throughput, delay
  bounds measured over some time period.  They may be bounds,
  statistical bounds, or percentiles (e.g., "90% of all packets
  measured over intervals of at least 5 minutes will cross the DS
  domain in less than 5 milliseconds").  A wide variety of
  characteristics may be used but they must be explicit, quantifiable,
  and defensible.  Where particular statistics are used, the document
  must be precise about how they are to be measured and about how the
  characteristics were derived.

  Advice to a network operator would be to use these as guidelines in
  creating a service specification rather than use them directly.  For
  example, a "loss-free" PDB would probably not be sold as such, but
  rather as a service with a very small packet loss probability.








Nichols & Carpenter          Informational                     [Page 14]

RFC 3086             Diffserv per Domain Behaviors            April 2001


5.4 Parameters

  The definition and characteristics of a PDB may be parameterized by
  network-specific features; for example, maximum number of hops,
  minimum bandwidth, total number of entry/exit points of the PDB
  to/from the diffserv network, maximum transit delay of network
  elements, minimum buffer size available for the PDB at a network
  node, etc.

5.5 Assumptions

  In most cases, PDBs will be specified assuming lossless links, no
  link failures, and relatively stable routing.  This is reasonable
  since otherwise it would be very difficult to quantify behavior and
  this is the operating conditions for which most operators strive.
  However, these assumptions must be clearly stated.  Since PDBs with
  specific bandwidth parameters require that bandwidth to be available,
  the assumptions to be stated may include standby capacity.  Some PDBs
  may be specifically targeted for cases where these assumptions do not
  hold, e.g., for high loss rate links, and such targeting must also be
  made explicit.  If additional restrictions, especially specific
  traffic engineering measures, are required, these must be stated.

  Further, if any assumptions are made about the allocation of
  resources within a diffserv network in the creation of the PDB, these
  must be made explicit.

5.6 Example Uses

  A PDB specification must give example uses to motivate the
  understanding of ways in which a diffserv network could make use of
  the PDB although these are not expected to be detailed.  For example,
  "A bulk handling PDB may be used for all packets which should not
  take any resources from the network unless they would otherwise go
  unused.  This might be useful for Netnews traffic or for traffic
  rejected from some other PDB by traffic policers."

5.7 Environmental Concerns (media, topology, etc.)

  Note that it is not necessary for a provider to expose which PDB (if
  a commonly defined one) is being used nor is it necessary for a
  provider to specify a service by the PDB's attributes.  For example,
  a service provider might use a PDB with a "no queueing loss"
  characteristic in order to specify a "very low loss" service.

  This section is to inject realism into the characteristics described
  above.  Detail the assumptions made there and what constraints that
  puts on topology or type of physical media or allocation.



Nichols & Carpenter          Informational                     [Page 15]

RFC 3086             Diffserv per Domain Behaviors            April 2001


5.8 Security Considerations for each PDB

  This section should include any security considerations that are
  specific to the PDB.  Is it subject to any unusual theft-of-service
  or denial-of-service attacks?  Are any unusual security precautions
  needed?

  It is not necessary to repeat the general security discussions in
  [RFC2474] and [RFC2475], but a reference should be included.  Also
  refer to any special security considerations for the PHB or PHBs
  used.

6 On PDB Attributes

  As discussed in section 4, measurable, quantifiable attributes
  associated with each PDB can be used to describe what will happen to
  packets using that PDB as they cross the domain.  In its role as a
  building block for the construction of interdomain quality-of-
  service, a PDB specification should provide the answer to the
  question: Under what conditions can we join the output of this domain
  to another under the same traffic conditioning and expectations?
  Although there are many ways in which traffic might be distributed,
  creating quantifiable, realizable PDBs that can be concatenated into
  multi-domain services limits the realistic scenarios.  A PDB's
  attributes with a clear statement of the conditions under which the
  attributes hold is critical to the composition of multi-domain
  services.

  There is a clear correlation between the strictness of the traffic
  conditioning and the quality of the PDB's attributes.  As indicated
  earlier, numerical bounds are likely to be statistical or expressed
  as a percentile.  Parameters expressed as strict bounds will require
  very precise mathematical analysis, while those expressed
  statistically can to some extent rely on experiment.  Section 7 gives
  the example of a PDB without strict traffic conditioning and
  concurrent work on a PDB with strict traffic conditioning and
  attributes is also in front of the WG [VW].  This section gives some
  general considerations for characterizing PDB attributes.

  There are two ways to characterize PDBs with respect to time.  First
  are properties over "long" time periods, or average behaviors.  A PDB
  specification should report these as the rates or throughput seen
  over some specified time period.  In addition, there are properties
  of "short" time behavior, usually expressed as the allowable
  burstiness in a traffic aggregate.  The short time behavior is
  important in understanding buffering requirements (and associated
  loss characteristics) and for metering and conditioning
  considerations at DS boundaries.  For short-time behavior, we are



Nichols & Carpenter          Informational                     [Page 16]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  interested primarily in two things: 1) how many back-to-back packets
  of the PDB's traffic aggregate will we see at any point (this would
  be metered as a burst) and 2) how large a burst of packets of this
  PDB's traffic aggregate can appear in a queue at once (gives queue
  overflow and loss).  If other PDBs are using the same PHB within the
  domain, that must be taken into account.

6.1 Considerations in specifying long-term or average PDB attributes

  To characterize the average or long-term behavior for the foo PDB we
  must explore a number of questions, for instance: Can the DS domain
  handle the average foo traffic flow?  Is that answer topology
  dependent or are there some specific assumptions on routing which
  must hold for the foo PDB to preserve its "adequately provisioned"
  capability?  In other words, if the topology of D changes suddenly,
  will the foo PDB's attributes change?  Will its loss rate
  dramatically increase?

  Let domain D in figure 4 be an ISP ringing the U.S. with links of
  bandwidth B and with N tails to various metropolitan areas.  Inside
  D, if the link between the node connected to A and the node connected
  to Z goes down, all the foo traffic aggregate between the two nodes
  must transit the entire ring: Would the bounded behavior of the foo
  PDB change?  If this outage results in some node of the ring now
  having a larger arrival rate to one of its links than the capacity of
  the link for foo's traffic aggregate, clearly the loss rate would
  change dramatically.  In this case, topological assumptions were made
  about the path of the traffic from A to Z that affected the
  characteristics of the foo PDB.  If these topological assumptions no
  longer hold, the loss rate of packets of the foo traffic aggregate
  transiting the domain could change; for example, a characteristic
  such as "loss rate no greater than 1% over any interval larger than
  10 minutes." A PDB specification should spell out the assumptions
  made on preserving the attributes.

                 ____X________X_________X___________          /
                /                                   \    L   |
        A<---->X                                     X<----->|  E
               |                                     |       |
               |               D                     |        \
        Z<---->X                                     |
               |                                     |
                \___________________________________/
                        X                 X

       Figure 4: ISP and DS domain D connected in a ring and
                 connected to DS domain E




Nichols & Carpenter          Informational                     [Page 17]

RFC 3086             Diffserv per Domain Behaviors            April 2001


6.2 Considerations in specifying short-term or bursty PDB attributes

  Next, consider the short-time behavior of the traffic aggregate
  associated with a PDB, specifically whether permitting the maximum
  bursts to add in the same manner as the average rates will lead to
  properties that aggregate or under what conditions this will lead to
  properties that aggregate.  In our example, if domain D allows each
  of the uplinks to burst p packets into the foo traffic aggregate, the
  bursts could accumulate as they transit the ring.  Packets headed for
  link L can come from both directions of the ring and back-to-back
  packets from foo's traffic aggregate can arrive at the same time.  If
  the bandwidth of link L is the same as the links of the ring, this
  probably does not present a buffering problem.  If there are two
  input links that can send packets to queue for L, at worst, two
  packets can arrive simultaneously for L.  If the bandwidth of link L
  equals or exceeds twice B, the packets won't accumulate.  Further, if
  p is limited to one, and the bandwidth of L exceeds the rate of
  arrival (over the longer term) of foo packets (required for bounding
  the loss) then the queue of foo packets for link L will empty before
  new packets arrive.  If the bandwidth of L is equal to B, one foo
  packet must queue while the other is transmitted.  This would result
  in N x p back-to- back packets of this traffic aggregate arriving
  over L during the same time scale as the bursts of p were permitted
  on the uplinks.  Thus, configuring the PDB so that link L can handle
  the sum of the rates that ingress to the foo PDB doesn't guarantee
  that L can handle the sum of the N bursts into the foo PDB.

  If the bandwidth of L is less than B, then the link must buffer
  Nxpx(B-L)/B foo packets to avoid loss.  If the PDB is getting less
  than the full bandwidth L, this number is larger.  For probabilistic
  bounds, a smaller buffer might do if the probability of exceeding it
  can be bounded.

  More generally, for router indegree of d, bursts of foo packets might
  arrive on each input.  Then, in the absence of any additional traffic
  conditioning, it is possible that dxpx(# of uplinks) back-to-back foo
  packets can be sent across link L to domain E.  Thus the DS domain E
  must permit these much larger bursts into the foo PDB than domain D
  permits on the N uplinks or else the foo traffic aggregate must be
  made to conform to the TCA for entering E (e.g., by shaping).

  What conditions should be imposed on a PDB and on the associated PHB
  in order to ensure PDBs can be concatenated, as across the interior
  DS domains of figure 1?  Traffic conditioning for constructing a PDB
  that has certain attributes across a DS domain should apply
  independently of the origin of the packets.  With reference to the





Nichols & Carpenter          Informational                     [Page 18]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  example we've been exploring, the TCA for the PDB's traffic aggregate
  entering link L into domain E should not depend on the number of
  uplinks into domain D.

6.3 Remarks

  This section has been provided as motivational food for thought for
  PDB specifiers.  It is by no means an exhaustive catalog of possible
  PDB attributes or what kind of analysis must be done.  We expect this
  to be an interesting and evolutionary part of the work of
  understanding and deploying differentiated services in the Internet.
  There is a potential for much interesting research work.  However, in
  submitting a PDB specification to the Diffserv WG, a PDB must also
  meet the test of being useful and relevant by a deployment
  experience, described in section 8.

7 A Reference Per-Domain Behavior

  The intent of this section is to define as a reference a Best Effort
  PDB, a PDB that has little in the way of rules or expectations.

7.1 Best Effort PDB

7.1.1 Applicability

  A Best Effort (BE) PDB is for sending "normal internet traffic"
  across a diffserv network.  That is, the definition and use of this
  PDB is to preserve, to a reasonable extent, the pre-diffserv delivery
  expectation for packets in a diffserv network that do not require any
  special differentiation.  Although the PDB itself does not include
  bounds on availability, latency, and packet loss, this does not
  preclude Service Providers from engineering their networks so as to
  result in commercially viable bounds on services that utilize the BE
  PDB.  This would be analogous to the Service Level Guarantees that
  are provided in today's single-service Internet.

  In the present single-service commercial Internet, Service Level
  Guarantees for availability, latency, and packet delivery can be
  found on the web sites of ISPs [WCG, PSI, UU].  For example, a
  typical North American round-trip latency bound is 85 milliseconds,
  with each service provider's site information specifying the method
  of measurement of the bounds and the terms associated with these
  bounds contractually.








Nichols & Carpenter          Informational                     [Page 19]

RFC 3086             Diffserv per Domain Behaviors            April 2001


7.1.2 TCS and PHB configurations

  There are no restrictions governing rate and bursts of packets beyond
  the limits imposed by the ingress link.  The network edge ensures
  that packets using the PDB are marked for the Default PHB (as defined
  in [RFC2474]), but no other traffic conditioning is required.
  Interior network nodes apply the Default PHB on these packets.

7.1.3 Attributes of this PDB

  "As much as possible as soon as possible".

  Packets of this PDB will not be completely starved and when resources
  are available (i.e., not required by packets from any other traffic
  aggregate), network elements should be configured to permit packets
  of this PDB to consume them.

  Network operators may bound the delay and loss rate for services
  constructed from this PDB given knowledge about their network, but
  such attributes are not part of the definition.

7.1.4 Parameters

  None.

7.1.5 Assumptions

  A properly functioning network, i.e., packets may be delivered from
  any ingress to any egress.

7.1.6 Example uses

  1. For the normal Internet traffic connection of an organization.

  2. For the "non-critical" Internet traffic of an organization.

  3. For standard domestic consumer connections

7.1.7 Environmental Concerns

  There are no environmental concerns specific to this PDB.

7.1.8 Security Considerations for BE PDB

  There are no specific security exposures for this PDB.  See the
  general security considerations in [RFC2474] and [RFC2475].





Nichols & Carpenter          Informational                     [Page 20]

RFC 3086             Diffserv per Domain Behaviors            April 2001


8 Guidelines for writing PDB specifications

  G1. Following the format given in this document, write a draft and
  submit it as an Internet Draft.  The document should have "diffserv"
  as some part of the name.  Either as an appendix to the draft, or in
  a separate document, provide details of deployment experience with
  measured results on a network of non-trivial size carrying realistic
  traffic and/or convincing simulation results (simulation of a range
  of modern traffic patterns and network topologies as applicable).
  The document should be brought to the attention of the diffserv WG
  mailing list, if active.

  G2. Initial discussion should focus primarily on the merits of the
  PDB, though comments and questions on the claimed attributes are
  reasonable.  This is in line with the Differentiated Services goal to
  put relevance before academic interest in the specification of PDBs.
  Academically interesting PDBs are encouraged, but would be more
  appropriate for technical publications and conferences, not for
  submission to the IETF.  (An "academically interesting" PDB might
  become a PDB of interest for deployment over time.)

  The implementation of the following guidelines varies, depending on
  whether there is an active diffserv working group or not.

  Active Diffserv Working Group path:

  G3. Once consensus has been reached on a version of a draft that it
  is a useful PDB and that the characteristics "appear" to be correct
  (i.e., not egregiously wrong) that version of the draft goes to a
  review panel the WG co-chairs set up to audit and report on the
  characteristics.  The review panel will be given a deadline for the
  review.  The exact timing of the deadline will be set on a case-by-
  case basis by the co-chairs to reflect the complexity of the task and
  other constraints (IETF meetings, major holidays) but is expected to
  be in the 4-8 week range.  During that time, the panel may correspond
  with the authors directly (cc'ing the WG co-chairs) to get
  clarifications.  This process should result in a revised draft and/or
  a report to the WG from the panel that either endorses or disputes
  the claimed characteristics.

  G4. If/when endorsed by the panel, that draft goes to WG last call.
  If not endorsed, the author(s) can give an itemized response to the
  panel's report and ask for a WG Last Call.








Nichols & Carpenter          Informational                     [Page 21]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  G5. If/when passes Last Call, goes to ADs for publication as a WG
  Informational RFC in our "PDB series".

  If no active Diffserv Working Group exists:

  G3. Following discussion on relevant mailing lists, the authors
  should revise the Internet Draft and contact the IESG for "Expert
  Review" as defined in section 2 of RFC 2434 [RFC2434].

  G4. Subsequent to the review, the IESG may recommend publication of
  the Draft as an RFC, request revisions, or decline to publish as an
  Informational RFC in the "PDB series".

9 Security Considerations

  The general security considerations of [RFC2474] and [RFC2475] apply
  to all PDBs.  Individual PDB definitions may require additional
  security considerations.

10 Acknowledgements

  The ideas in this document have been heavily influenced by the
  Diffserv WG and, in particular, by discussions with Van Jacobson,
  Dave Clark, Lixia Zhang, Geoff Huston, Scott Bradner, Randy Bush,
  Frank Kastenholz, Aaron Falk, and a host of other people who should
  be acknowledged for their useful input but not be held accountable
  for our mangling of it.  Grenville Armitage coined "per domain
  behavior (PDB)" though some have suggested similar terms prior to
  that.  Dan Grossman, Bob Enger, Jung-Bong Suk, and John Dullaert
  reviewed the document and commented so as to improve its form.

References

  [RFC2474] Nichols, K., Blake, S. Baker, F. and D. Black, "Definition
            of the Differentiated Services Field (DS Field) in the IPv4
            and IPv6 Headers", RFC 2474, December 1998.

  [RFC2475] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z. and
            W. Weiss, "An Architecture for Differentiated Services",
            December 1998.

  [RFC2597] Heinanen, J., Baker, F., Weiss, W. and J. Wroclawski,
            "Assured Forwarding PHB Group", RFC 2597, June 1999.

  [RFC2598] Jacobson, V., Nichols, K. and K. Poduri, "An Expedited
            Forwarding PHB", RFC 2598, June 1999.





Nichols & Carpenter          Informational                     [Page 22]

RFC 3086             Diffserv per Domain Behaviors            April 2001


  [RFC2698] Heinanen, J. and R. Geurin, "A Two Rate Three Color
            Marker", RFC 2698, June 1999.

  [MODEL]   Bernet, Y., Blake, S., Grossman, D. and A. Smith, "An
            Informal Management Model for Diffserv Routers", Work in
            Progress.

  [MIB]     Baker, F., Chan, K. and A. Smith, "Management Information
            Base for the Differentiated Services Architecture", Work in
            Progress.

  [VW]      Jacobson, V., Nichols, K. and K. Poduri, "The 'Virtual
            Wire' Per-Domain Behavior", Work in Progress.

  [WCG]     Worldcom, "Internet Service Level Guarantee",
            http://www.worldcom.com/terms/service_level_guarantee/
            t_sla_terms.phtml

  [PSI]     PSINet, "Service Level Agreements",
            http://www.psinet.com/sla/

  [UU]      UUNET USA Web site, "Service Level Agreements",
            http://www.us.uu.net/support/sla/

  [RFC2434] Alvestrand, H. and T. Narten, "Guidelines for IANA
            Considerations", BCP 26, RFC 2434, October 1998.

Authors' Addresses

  Kathleen Nichols
  Packet Design, LLC
  2465 Latham Street, Third Floor
  Mountain View, CA 94040
  USA

  EMail: [email protected]


  Brian Carpenter
  IBM
  c/o iCAIR
  Suite 150
  1890 Maple Avenue
  Evanston, IL 60201
  USA

  EMail: [email protected]




Nichols & Carpenter          Informational                     [Page 23]

RFC 3086             Diffserv per Domain Behaviors            April 2001


Full Copyright Statement

  Copyright (C) The Internet Society (2001).  All Rights Reserved.

  This document and translations of it may be copied and furnished to
  others, and derivative works that comment on or otherwise explain it
  or assist in its implementation may be prepared, copied, published
  and distributed, in whole or in part, without restriction of any
  kind, provided that the above copyright notice and this paragraph are
  included on all such copies and derivative works.  However, this
  document itself may not be modified in any way, such as by removing
  the copyright notice or references to the Internet Society or other
  Internet organizations, except as needed for the purpose of
  developing Internet standards in which case the procedures for
  copyrights defined in the Internet Standards process must be
  followed, or as required to translate it into languages other than
  English.

  The limited permissions granted above are perpetual and will not be
  revoked by the Internet Society or its successors or assigns.

  This document and the information contained herein is provided on an
  "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
  TASK FORCE DISCLAIMS 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.

Acknowledgement

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



















Nichols & Carpenter          Informational                     [Page 24]