[Search] [txt|pdf|bibtex] [Tracker] [WG] [Email] [Diff1] [Diff2] [Nits]

Versions: 00 01 02 03 rfc2474                                           
INTERNET-DRAFT                                          Kathleen Nichols
Diffserv Working Group                                      Bay Networks
Expires: January 1999                                       Steven Blake
                                         Torrent Networking Technologies
                                                              Fred Baker
                                                           Cisco Systems
                                                          David L. Black
                                                          The Open Group

                                                               July 1998


        Definition of the Differentiated Services Field (DS Field)
                      in the IPv4 and IPv6 Headers

                    <draft-ietf-diffserv-header-01.txt>


Status of This Memo

   This document is an Internet-Draft.  Internet-Drafts are working
   documents of the Internet Engineering Task Force (IETF), its areas,
   and its working groups.  Note that other groups may also distribute
   working documents as Internet-Drafts.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   To view the entire list of current Internet-Drafts, please check the
   "1id-abstracts.txt" listing contained in the Internet-Drafts Shadow
   Directories on ftp.is.co.za (Africa), ftp.nordu.net (Northern
   Europe), ftp.nis.garr.it (Southern Europe), munnari.oz.au (Pacific
   Rim), ftp.ietf.org (US East Coast), or ftp.isi.edu (US West Coast).

Copyright Notice

   Copyright (C) The Internet Society (1998).  All Rights Reserved.

Abstract

   Differentiated services are intended to provide scalable service
   discrimination in the Internet without the need for per-flow state
   and signaling at every hop.  A variety of services may be built from
   a small, well-defined set of building blocks which are deployed in
   network nodes.  The services may be either end-to-end or intra-
   domain.  Services can be constructed by a combination of:

   - setting bits in an IP header field at network boundaries and
     administrative boundaries,
   - using those bits to determine how packets are forwarded by the
     nodes inside the network, and
   - conditioning the marked packets at network boundaries in accordance
     with the requirements or rules of each service.

   The requirements or rules of each service must be set through
   administrative policy mechanisms which are outside the scope of this
   document.  A differentiated services-compliant network node includes a
   classifier that selects packets based on the value of the DS field
   and is capable of delivering the specific packet forwarding treatment
   indicated by the DS field value.  Setting of the DS field and
   conditioning of the temporal behavior of marked packets need only be
   performed at network boundaries and may vary in complexity.

   This document defines the IP header field, called the DS (for
   differentiated services) field.  In IPv4, it defines the layout of
   the TOS octet; in IPv6, the Traffic Class octet.  In addition, a base
   set of packet forwarding treatments, or per-hop behaviors, is
   defined.

   For a more complete understanding of differentiated services, see
   also the differentiated services architecture [ARCH].


1.  Introduction

   Differentiated services are intended  to provide a framework and
   building blocks to enable deployment of scalable service
   discrimination in the Internet.  The differentiated services approach
   aims to speed deployment by separating the architecture into two
   major components, one of which is fairly well-understood and the
   other of which is just beginning to be understood.  In this, we are
   guided by the original design of the Internet where the decision was
   made to separate the forwarding and routing components.  Packet
   forwarding is the relatively simple task that needs to be done on a
   per-packet basis as quickly as possible.  Forwarding uses the packet
   header to find an entry in a routing table that determines the
   packet's output interface.  Routing sets the entries in that table
   and may need to reflect a range of transit and other policies as well
   as to keep track of route failures.  Routing tables are maintained as
   a background process to the forwarding task.  Further, routing is the
   more complex task and it has continued to evolve over the past 20
   years.

   Analogously, the differentiated services architecture contains two
   main components.  One is the fairly well-understood behavior in the
   forwarding path and the other is the more complex and still emerging
   background policy and allocation component that configures parameters
   used in the forwarding path.  The forwarding path behaviors include
   the differential treatment an individual packet receives, as
   implemented by queueing service disciplines and/or queue management
   disciplines.  These per-hop behaviors are useful and required in
   network nodes to deliver differentiated treatment of packets no
   matter how we construct end-to-end or intra-domain services.  These
   behaviors and the mechanisms to select them on a per-packet basis can
   be deployed in network nodes today and it is this aspect of the
   differentiated services architecture that is being addressed first.
   In addition, the forwarding path may require that some monitoring,
   policing, and shaping be done on the network traffic designated for
   "special" treatment in order to enforce requirements associated with
   the delivery of the special treatment.  Mechanisms for this kind of
   traffic conditioning are also fairly well-understood.  The wide
   deployment of such traffic conditioners is also important to enable
   the construction of services, though their actual use in constructing
   services may evolve over time.

   The configuration of network elements with respect to which packets
   get special treatment and what kinds of rules are to be applied to
   the use of resources is much less well-understood.  Nevertheless,
   it is possible to deploy useful differentiated services in networks
   by using simple policies and static configurations.  As described in
   [ARCH], there are a number of ways to compose per-hop behaviors and
   traffic conditioners to create services.  In the process, additional
   experience is gained that will guide more complex policies and
   allocations.  The basic behaviors in the forwarding path can remain
   the same while this component of the architecture evolves.
   Experiences with the construction of such services will continue for
   some time, thus we avoid standardizing this construction as
   premature.  Further, much of the details of service construction are
   covered by legal agreements between different business entities and
   we avoid this as very much outside the scope of the IETF.

   This document concentrates on the forwarding path component.  In the
   packet forwarding path, differentiated services are realized by
   mapping the codepoint contained in a field in the IP packet header to
   a particular forwarding treatment, or per-hop behavior (PHB), at each
   network node along its path.  The codepoints may be chosen from a set
   of recommended values or may have purely local meaning.  PHBs are
   expected to be implemented by employing a range of queue service and/
   or queue management disciplines on a network node's output interface
   queue, for example weighted round-robin queue servicing or drop-
   preference queue management.

   Marking is performed by traffic conditioners at network boundaries,
   including the edges of the network (first hop router or source host)
   and administrative boundaries.  Traffic conditioners may include the
   primitives of marking, metering, policing and shaping (these
   mechanisms are described in [ARCH]).  Services are realized by the
   use of particular packet classification and traffic conditioning
   mechanisms at boundaries coupled with the concatenation of per-hop
   behaviors along the transit path of the traffic.  A goal of the
   differentiated services architecture is to specify these building
   blocks for future extensibility, both of the number and type of the
   building blocks and of the services built from them.

   Terminology used in this draft is defined in Sec. 2.  The
   differentiated services field definition (DS field) is given in Sec.
   3.  In Sec. 4, we discuss the desire for partial backwards
   compatibility with current use of the IP precedence field. As a
   solution, we introduce Class Selector codepoints and Class Selector
   Compliant PHBs.  Sec. 5 presents guidelines for per-hop behavior
   standardization.  Sec. 6 discusses guidelines for allocation of
   codepoints.  Sec. 7 covers security considerations.  We present
   examples of class selector compliant PHB implementations in the
   Appendix.

   This document is a concise description of the DS field and its uses.
   It is intended to be read along with the differentiated services
   architecture [ARCH].

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED",  "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in [RFC2119].


2.  Terminology Used in This Document

   Behavior aggregate: a collection of packets with the same codepoint
   crossing a boundary in a particular direction.  The terms "aggregate"
   and "behavior aggregate" are used interchangeably in this document.

   Boundary: the edge of a DS domain, where classifiers and traffic
   conditioners are likely to be deployed.  A 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 boundary typically is found at the ingress to
   the first-hop differentiated services-compliant router (or network
   node) a host's packets traverse, or at the egress of the last-hop
   differentiated services-compliant router or network node packets
   traverse before arriving at a host.  This is sometimes referred to as
   the boundary at a leaf router.  A boundary might be co-located with a
   host, subject to local policy.

   Classifier: an entity which selects packets based on the content of
   packet headers according to defined rules.

   Codepoint: a specific value of the DSCP portion of the DS field.
   Recommended codepoints SHOULD map to specific, standardized PHBs.
   Multiple codepoints MAY map to the same PHB.

   Differentiated Service-compliant: in compliance with the requirements
   specified in this document.

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

   Differentiated Services field: the IPv4 header TOS octet or the
   IPv6 Traffic Class octet when interpreted in conformance with the
   definition given in this document.  Also DS field.

   Mechanism:  The implementation of a per-hop behavior according to a
   particular algorithm.

   Microflow: a single instance of an application-to-application flow of
   packets which is identified by source address, source port,
   destination address, destination port and protocol id.

   Per-hop Behavior (PHB): a description of the externally observable
   forwarding treatment applied at a differentiated services-compliant
   node to a behavior aggregate.  The description of a PHB should
   be sufficiently detailed to allow the construction of predictable
   services.

   Traffic conditioning: control functions that can be applied to a
   behavior aggregate, application flow, or other operationally useful
   subset of traffic, e.g., routing updates.  These may include
   metering, policing, shaping, and packet marking.  Traffic
   conditioning is used to enforce service level agreements between
   domains and to condition traffic to receive a differentiated service
   within a domain by marking packets with the appropriate codepoint in
   the DS field and by monitoring and altering the temporal
   characteristics of the aggregate where necessary.

   Traffic conditioner: an entity which performs traffic conditioning
   functions and which may contain meters, policers, shapers, and
   markers.  Traffic conditioners are typically deployed in boundary
   nodes only.

   Service: a description of the overall treatment of a customer's
   traffic within a particular domain, within a set of interconnected
   DS domains, or end-to-end.  Service descriptions are covered by
   administrative policy and services are constructed by applying
   traffic conditioning to create behavior aggregates which experience a
   known PHB at each hop within the DS domain.  Multiple services can be
   supported by a single per-hop behavior used in concert with a range
   of traffic conditioners.

   To summarize, classifiers and traffic conditioners are used to select
   which packets are to be added to behavior aggregates. Aggregates receive
   differentiated treatment in a domain and may alter the temporal
   characteristics of the aggregate to conform to some requirements.  A
   packet's DS field is used to designate the packet's behavior
   aggregate and is subsequently used to determine which forwarding
   treatment the packet receives.  A DS field classifier which can
   select a differential output queue servicing discipline (or PHB)
   based on the codepoint in the DS field SHOULD be in all network
   nodes of a differentiated services domain.  The classifiers
   and traffic conditioners are configured in accordance with some
   service specification, a matter of administrative policy outside the
   scope of this document.

   Additional differentiated services definitions are in [ARCH].


3.  Differentiated Services Field Definition

   A new header field, called the DS field, is defined, which is
   intended to supersede the existing definitions of the IPv4 TOS octet
   [RFC791] and the IPv6 Traffic Class octet [IPv6].

   Six bits of the DS field are used as a codepoint (DSCP) to select the
   PHB a packet experiences at each node.  A two-bit currently unused
   (CU) field is reserved and may be assigned later, e.g., for explicit
   congestion notification.  The value of the CU bits MUST be
   ignored by differentiated services-compliant nodes when determining
   the per-hop behavior to apply to a received packet.

   The DS field structure is presented below:


        0   1   2   3   4   5   6   7
      +---+---+---+---+---+---+---+---+
      |         DSCP          |  CU   |
      +---+---+---+---+---+---+---+---+

        DSCP: differentiated services codepoint
        CU:   currently unused


   Implementors should note that the DSCP field is six bits wide.  DS-
   compliant nodes MUST select PHBs by matching against the entire 6-
   bit DSCP field, e.g., by treating the value of the field as a table
   index which is used to select a particular packet handling mechanism
   which has been implemented in that device.  The DSCP field is defined
   as an unstructured field to facilitate the definition of future per-
   hop behaviors.

   With some exceptions noted below, the mapping of codepoints to PHBs
   MUST be configurable.  A DS-compliant node MUST support the logical
   equivalent of a configurable mapping table from codepoints to PHBs.
   PHB specifications MUST include a recommended default codepoint, but
   operators MAY choose to use different codepoints, either in addition
   to or in place of the recommended default.  Note that if operators do
   so choose, re-marking of DS fields will be necessary at
   administrative boundaries even if the same PHBs are implemented on
   both sides of the boundary.  See [ARCH] for further discussion of re-
   marking.  The recommended default codepoint for a PHB must be unique
   for codepoints in the standard space (see Sec. 6).

   The exceptions to configurability are for codepoints 'xxx000' and are
   noted in Sec. 4.

   Packets received with an unrecognized codepoint SHOULD be forwarded
   as if they were marked for the Default behavior (see Sec. 4).  Such
   packets MUST NOT cause the network node to crash.

   The structure of the DS field shown above is incompatible with the
   existing definition of the IPv4 TOS octet in [RFC791, RFC1349].  The
   presumption is that differentiated services domains protect
   themselves by deploying re-marking boundary nodes, as should networks
   using the RFC 791 Precedence designations.  Correct operational
   procedure should follow [RFC791], which states: "If the actual use of
   these precedence designations is of concern to a particular network,
   it is the responsibility of that network to control the access to,
   and use of, those precedence designations."  Validating the value of
   the DS field at network boundaries is sensible in any case since an
   upstream node can easily set it to any random value.  Differentiated
   services domains which are not suitably isolated by a re-marking
   boundary node may deliver unpredictable service.

   Nodes MAY rewrite the DS field as needed to provide a desired local
   or end-to-end service.  Specifications of DS field translations at
   network boundaries are the subject of Service Level Agreements
   between providers and users, and are outside the scope of this
   document.  Standardized PHBs allow providers to build their services
   from a well-known set of packet forwarding treatments that can be
   expected to be present in the equipment of many vendors.


4.  Historical Codepoint Definitions and PHB Requirements

   The DS field will have a limited backwards compatibility with current
   practice, described in this section.  Backwards compatibility is
   addressed in two ways.  First, per-hop behaviors that are
   already in widespread use MUST be included in a DS-compliant
   node.  In addition, there are some codepoints that correspond to
   historical use of the IP Precedence field and we reserve these
   codepoints to map to PHBs that meet the general requirements
   specified in [RFC791], though the specific differentiated services
   PHBs mapped to by those codepoints MAY have additional
   specifications.


4.1  A Default PHB

   A "default" PHB MUST be available in a DS-compliant node.  This is
   the common, best-effort forwarding behavior available in existing
   routers as standardized in [RFC1812].  When no other agreements are
   in place, it is assumed that packets belong to this aggregate.  Such
   packets may be sent into a network without adhering to any particular
   rules and the network will deliver as many of these packets as
   possible and as soon as possible, subject to other resource policy
   constraints.  A reasonable implementation of this PHB would be a
   queueing discipline that sends packets of this aggregate whenever the
   output link is not required to satisfy another PHB.  A reasonable
   policy for constructing services would ensure that the aggregate was
   not "starved" or else the network provider might become quite
   unpopular.  This permits senders that are not differentiated
   services-aware to continue to use the network in the same manner as
   today and to have the same kinds of expectations from the network
   regarding service.  The RECOMMENDED codpoint for the Default PHB is
   the bit pattern '000000'; the value '000000' MUST map to a PHB that
   meets these specifications.  The codepoint chosen for Default
   behavior is compatible with existing practice [RFC791, RFC1349].

   A packet initially marked for the Default behavior MAY be re-marked
   with another codepoint as it passes a boundary into another DS domain
   so that it will be forwarded using a different PHB within the domain,
   possibly subject to some negotiated agreement between the peering
   domains.

4.2  Once and Future IP Precedence Field Use

   We wish to maintain some form of backward compatibility with present
   uses of the IP Precedence Field: bits 0-2 of the IPv4 TOS octet.
   Routers exist that use the IP Precedence field to select different
   per-hop forwarding treatments quite similarly to the use proposed
   here for the DSCP field.  Thus, a simple prototype diffserv
   architecture can be quickly deployed by appropriately configuring
   these routers.  Further, IP systems today understand the location of
   the IP Precedence field, and thus if these bits are used in a similar
   manner as DS-compliant equipment is deployed, significant failures
   are not likely during early deployment.  In other words, strict DS-
   compliance need not be ubiquitous even within a single service
   provider's network if bits 0-2 of the DSCP field are employed in a
   manner similar to, or subsuming, the deployed uses of the IP
   Precedence field.

4.2.1  IP Precedence History and Evolution in Brief

   The IP Precedence field is something of a forerunner of the DS field.
   IP Precedence, and the IP Precedence Field, were first defined in
   [RFC791].  The values that the three-bit IP Precedence Field might
   take were assigned to various uses, including network control
   traffic, routing traffic, and various levels of privilege.  The least
   level of privilege was deemed "routine traffic".  In [RFC791], the
   notion of Precedence was defined broadly as " An independent measure
   of the importance of this datagram."  Not all values of the IP
   Precedence field were assumed to have meaning across boundaries, for
   instance "The Network Control precedence designation is intended to
   be used within a network only.  The actual use and control of that
   designation is up to each network." [RFC791]

   Although early BBN IMPs implemented the service, early commercial
   routers and UNIX IP forwarding code generally did not.  As networks
   became more complex and customer requirements grew, commercial
   routers developed ways to implement various kinds of queueing
   services including priority queueing, which were generally based on
   policies encoded in filters in the routers, which looked at IP
   addresses, IP protocol numbers, TCP or UDP ports, and other header
   fields.  IP Precedence was and is among the options such filters can
   look at.

   In short, IP Precedence is widely deployed and widely used, if not in
   exactly the manner intended in [RFC791].  This was recognized in
   [RFC1122], which states that while the use of the IP Precedence field
   is valid, the specific assignment of the priorities in [RFC791] were
   merely historical.

4.2.2  Subsuming IP Precedence into Class Selector Codepoints

   A specification of the packet forwarding treatments selected by the
   IP precedence field today would have to be quite general; probably
   not specific enough to build predictable services from in the
   differentiated services framework.  To preserve partial backwards
   compatibility with known current uses of the IP Precedence field
   without sacrificing future flexibility, we have taken the approach of
   describing minimum requirements on a set of PHBs that are compatible
   with most of the existing forwarding treatments selected by the IP
   Precedence field.  In addition, we give a set of codepoints that MUST
   map to PHBs meeting minimum requirements.  The PHBs mapped to by
   these codepoints MAY have a more detailed list of specifications in
   addition to the required ones stated here.  Other codepoints MAY map
   to the same PHBs.  We refer to this set of codepoints as the Class
   Selector codepoints and the minimum requirements for PHBs that these
   codepoints may map to are called the Class Selector PHB requirements.

   4.2.2.1  The Class Selector Codepoints

   The DS field values of 'xxx000|xx', or DSCP = 'xxx000' and CU
   subfield unspecified are reserved as a set of Class Selector Codepoints.
   PHBs which are mapped to by these codepoints MUST meet the Class
   Selector PHB requirements in addition to preserving the Default PHB
   requirement on codepoint '000000'.

   4.2.2.2  The Class Selector PHB Requirements

   We refer to a Class Selector codepoint with a larger numerical value
   than another Class Selector codepoint as having a higher relative
   order while a Class Selector codepoint with a smaller numerical value
   than another Class Selector codepoint is said to have a lower
   relative order.  PHBs selected by a Class Selector codepoint SHOULD
   give packets a probability of timely forwarding that is not lower
   than that given to packets marked with a Class Selector codepoint of
   lower relative order.  Although a particular packet marked with a
   lower relative order Class Selector codepoint might be observed to be
   forwarded earlier than a particular packet marked with a higher
   relative order, sample observations that take place over a reasonable
   period of network history will conform to this requirement.  A
   discarded packet is considered to be an extreme case of untimely
   forwarding.

   Further, PHBs selected by distinct Class Selector codepoints SHOULD
   be independently forwarded; that is, packets marked with different
   Class Selector codepoints MAY be reordered.  A network node MAY have
   limits on the amount of the node's resources that can be used by each
   of these PHBs.

   PHB groups whose specification meets these requirements are referred
   to as Class Selector Compliant PHBs.

   It is easy to see that the Class Selector PHB Requirements on
   codepoint '000000' are compatible with those listed in Sec. 4.1
   for the Default PHB.

   4.2.2.3  Using the Class Selector PHB Requirements for IP Precedence
            Compatibility

   A DS-compliant network node can be deployed with a set of one or more
   Class Selector Compliant PHB groups.  This document states that the
   set of codepoints 'xxx000' MUST map to such a set of PHBs.  As it is
   also possible to map multiple codepoints to the same PHB, the vendor
   or the network administrator might configure the network node to map
   codepoints to PHBs irrespective of bits 3-5 of the DSCP field to
   yield a network that is compatible with historical IP Precedence use.
   Thus, for example, codepoint '011010' would map to the same PHB as
   codepoint '011000'.

   4.2.2.4  Example Mechanisms for Implementing Class Selector Compliant
            PHB Groups

   Class Selector Compliant PHBs can be realized by a variety of
   mechanisms, including strict priority queueing, WFQ or variants
   [HPFQA], weighted round-robin queueing (WRR), or Class-Based Queuing
   [CBQ].  The distinction between PHBs and mechanisms is described in
   more detail in Sec. 5.

   It is important to note that these mechanisms might be available
   through other PHBs (standardized or not) that are available in a
   particular vendor's equipment.  For example, future documents may
   standardize a Strict Priority Queueing PHB for a set of recommended
   codepoints.  A network administrator might configure those routers to
   select the Strict Priority Queueing PHBs with codepoints 'xxx000' in
   conformance with the requirements of this document.

   As a further example, another vendor might employ a CBQ mechanism in
   its routers.  The CBQ mechanism could be used to implement the Strict
   Priority Queueing PHBs as well as a set of Class Selector Compliant
   PHBs with a wider range of features than would be availble in a set
   of PHBs that did no more than meet the minimum Class Selector PHB
   requirements.  More examples are given in the Appendix.


4.3 Summary

   This document defines codepoints 'xxx000' as the Class Selector
   codepoints, where PHBs selected by these codepoints must meet the
   Class Selector PHB Requirements described in Sec. 4.2.2.2.  This is
   done to preserve a useful level of backward compatibility with
   current uses of the IP Precedence field in the Internet without
   unduly limiting future flexibility.  In addition, codepoint '000000'
   is used as the Default PHB value for the Internet and, as such, is
   not configurable.  The remaining seven non-zero Class Selector
   codepoints are configurable only to the extent that they map to PHBs
   that meet the requirements in Sec. 4.2.2.2.


5.  Per-Hop Behavior Standardization Guidelines

   Thirty-two DS codepoints are reserved for standardization as
   RECOMMENDED codepoints, and 32 codepoints are reserved for
   experimental or local use (EXP/LU) (see Sec. 6 for further details).

   Codepoint space          Usage Policy
   ---------------          ------------

     xxxxx0                 Reserved for RECOMMENDED codepoints
     xxxxx1                 EXP/LU

   The behavioral characteristics of a PHB are to be standardized, and
   not the algorithms or the mechanisms used to implement them.  A
   node may have a (possibly large) set of parameters that can be used
   to control how packets are scheduled onto an output interface (e.g.,
   N separate queues with settable priorities, queue lengths, round-
   robin weights, drop algorithm, drop preference weights and
   thresholds, etc).  To illustrate the distinction between a PHB and a
   mechanism, we point out that Class Selector Compliant PHBs might be
   implemented by several mechanisms, including: strict priority
   queueing combined with WFQ or variants [HPFQA], weighted round-robin
   queueing, or CBQ [CBQ].

   It is RECOMMENDED that PHB implementations do not introduce any
   packet reordering within a microflow.  Where PHBs are defined as a
   group, their definitions MUST specify any possible packet re-ordering
   implications which may occur if different packets within a microflow
   are marked for different PHBs within the group.

   Network equipment vendors are free to offer whatever parameters and
   capabilities are deemed useful or marketable.  When a particular,
   standardized PHB is implemented in a node, a vendor may use any
   algorithm that satisfies the definition of the PHB according to the
   standard.  The node's capabilities and its particular configuration
   determine the different ways that packets can be treated.

   Service providers are not required to use the same node mechanisms
   or configurations to enable service differentiation within their
   networks, and are free to configure the node parameters in whatever
   way that is appropriate for their service offerings and traffic
   engineering objectives.  Over time certain common per-hop behaviors
   are likely to evolve (i.e., ones that are particularly useful for
   implementing end-to-end services) and these may be associated with
   particular EXP/LU PHB codepoints in the DS field, allowing use across
   domain boundaries (see Sec. 6).  These PHBs are candidates for future
   standardization.

   Only those per-hop behaviors that are not described by an existing
   PHB standard, and have been implemented, deployed, and shown to be
   useful, should be standardized.  Since current experience with
   differentiated services is quite limited, it is premature to
   hypothesize the exact specification of these per-hop behaviors.  This
   specification has left room in the codepoint space to allow for
   evolution, thus the defined space is intentionally sparse.


6.  IANA Considerations

   The DSCP field in the DS field is capable of conveying 64 distinct
   codepoints.  The codepoint space is divided into three pools for the
   purpose of codepoint assignment and management: a pool of 32
   RECOMMENDED codepoints (Pool 1) to be assigned by Standards Action as
   defined in [CONS], a pool of 16 codepoints (Pool 2) to be reserved for
   experimental or Local Use (EXP/LU) as defined in [CONS], and a pool
   of 16 codepoints (Pool 3) which are initially available for
   experimental or local use, but which should be preferentially
   utilized for standardized assignments if Pool 1 is ever exhausted.
   The pools are defined in the following table (where 'x' refers to
   either '0' or '1'):


   Pool        Codepoint space          Assignment Policy
   ----        ---------------          -----------------

    1            xxxxx0                 Standards Action
    2            xxxx11                 EXP/LU
    3            xxxx01                 EXP/LU (*)

   (*) may be utilized for future Standards Action allocations as
       necessary


   This document defines the assignment of eight codepoints (xxx000)
   which are drawn from Pool 1 above.  These codepoints MUST be
   mapped, not to specific PHBs, but to PHBs that meet "at least" the
   requirements set out in Sec. 4.2.2.2 to provide a minimal level of
   backwards compatibility with IP Precedence as defined in [RFC791] and
   as deployed in some current equipment.

7.  Security Considerations

   This section considers security issues raised by the introduction of
   differentiated services, primarily the potential for denial-of-
   service attacks, and the related potential for theft of service by
   unauthorized traffic (Section 7.1).  Section 7.2 addresses the operation
   of differentiated services in the presence of IPsec including its
   interaction with IPsec tunnel mode and other tunnelling protocols.
   More extensive treatment of the security concerns raised by the overall
   differentiated services architecture are discussed in [ARCH].

7.1  Theft and Denial of Service

   The primary goal of differentiated services is to allow different
   levels of service to be provided for traffic streams on a common
   network infrastructure.  A variety of techniques may be used to
   achieve this, but the end result will be that some packets receive
   different (e.g., better) service than others.  The mapping of network
   traffic to the specific behaviors that result in different (e.g.,
   better or worse) service is indicated primarily by the DS codepoint,
   and hence an adversary may be able to obtain better service by
   modifying the codepoint to values indicating behaviors used for
   enhanced services or by injecting packets with such codepoint values.
   Taken to its limits, such theft of service becomes a denial-of-service
   attack when the modified or injected traffic depletes the resources
   available to forward it and other traffic streams.

   The defense against this class of theft- and denial-of-service attacks
   consists of the combination of traffic conditioning at network boundaries
   with security and integrity of the network infrastructure within a
   DS domain.  Network boundary nodes MUST ensure that all traffic entering
   the domain has codepoint values appropriate to the traffic and the domain,
   remarking the traffic with new codepoint values if necessary.
   These boundary nodes are the primary line of defense against theft-
   and denial-of-service attacks based on modified codepoints, as
   success of any such attack indicates that the codepoints used
   by the attacking traffic were inappropriate.  An important instance of a
   boundary node is that any traffic-originating node in a DS domain
   is the initial boundary node for that traffic.  Interior nodes
   in a DS domain rely on DS codepoints to associate traffic with the
   forwarding PHBs, and are not required to check codepoint values before
   using them.  As a result, the interior nodes depend on the correct
   operation of the DS domain to prevent the arrival of traffic with
   inappropriate codepoints that would disrupt operation of the domain.

7.2  IPsec and Tunnelling Interactions

   The IPsec protocol, as defined in [ESP,AH], does not include the IP
   header's DS field in any of its cryptographic calculations (in the
   case of tunnel mode, it is the outer IP header's DS field that is not
   included).  Hence modification of the DS field by a network node has
   no effect on IPsec's end-to-end security, because it cannot cause any
   IPsec integrity check to fail.  As a consequence, IPsec does not
   provide any defense against an adversary's modification of the DS
   field (i.e., a man-in-the-middle attack), as the adversary's
   modification will also have no effect on IPsec's end-to-end security.

   IPsec's tunnel mode provides security for the encapsulated IP
   header's DS field.  A tunnel mode IPsec packet contains two IP
   headers: an outer header supplied by the tunnel ingress node and an
   encapsulated inner header supplied by the original source of the
   packet.  When an IPsec tunnel is hosted (in whole or in part) on a
   differentiated services network, the intermediate network nodes
   operate on the DS field in the outer header.  At the tunnel egress
   node, IPsec processing includes removing the outer header and
   forwarding the packet (if required) using the inner header.
   The IPsec protocol REQUIRES that the inner header's DS field not
   be changed by this decapsulation processing to ensure that modifications
   to the DS field cannot be used to launch theft- or denial-of-service
   attacks across an IPsec tunnel endpoint.  This document makes no
   change to that requirement.  If the inner IP header has not been
   processed by a boundary node for the tunnel egress node's DS domain,
   the tunnel egress node is the boundary node for traffic exiting the
   tunnel, and hence MUST ensure that the resulting traffic has
   appropriate DS codepoints.

   When IPsec tunnel egress decapsulation processing includes a
   sufficiently strong cryptographic integrity check of the encapsulated
   packet (where sufficiency is determined by local security policy), the
   tunnel egress node can safely assume that the DS field in the inner
   header has the same value as it had at the tunnel ingress node.
   An important consequence is that otherwise insecure links
   within a DS domain can be secured by a sufficiently strong
   IPsec tunnel.  This analysis and its implications apply to any
   tunneling protocol that performs integrity checks, but the level
   of assurance of the inner header's DS field depends on the strength
   of the integrity check performed by the tunneling protocol.  In
   the absence of sufficient assurance for a tunnel that may transit
   nodes outside the current DS domain (or is otherwise vulnerable),
   the encapsulated packet MUST be treated as if it had arrived at a
   boundary from outside the DS domain.

8.  Acknowledgements

   The authors would like to acknowledge the Differentiated Services
   Working Group for discussions which helped shape this document.


9.  References

   [ACTIVE]    B. Braden, et. al., "Recommendations on Queue Management
               and Congestion Avoidance in the Internet", Internet RFC
               2309, April 1998.

   [AH]        S. Kent and R. Atkinson, "IP Authentication Header",
               Internet Draft <draft-ietf-ipsec-auth-header-07.txt>,
               July 1998.

   [ARCH]      Differentiated Services Architecture Document (work in
               preparation).

   [CBQ]       S. Floyd and V. Jacobson, "Link-sharing and Resource
               Management Models for Packet Networks", IEEE/ACM
               Transactions on Networking, Vol. 3 no. 4, pp. 365-386,
               August 1995.

   [CLARK]     D. Clark and W. Fang, "Explicit Allocation of Best
               Effort Packet Delivery Service",
               http://diffserv.lcs.mit.edu/Papers/exp-alloc-ddc-wf.pdf

   [CONS]      T. Narten and H. Alvestrand, "Guidelines for Writing an
               IANA Considerations Section in RFCs", Internet Draft
               <draft-iesg-iana-considerations-03.txt>, March 1998.

   [CCBES]     C. Lefelhocz, B. Lyles, S. Shenker, and L. Zhang,
               "Congestion Control for Best-Effort Service: Why We Need
               a New Paradigm", IEEE Network, Vol. 10, no. 1, January
               1996.

   [ESP]       S. Kent and R. Atkinson, "IP Encapsulating Security
               Payload (ESP)", Internet Draft
               <draft-ietf-ipsec-esp-v2-06.txt>, July 1998.

   [HPFQA]     J. Bennett and Hui Zhang, "Hierarchical Packet Fair
               Queueing Algorithms", Proc. ACM SIGCOMM 96, August 1996.

   [IPv6]      S. Deering and R. Hinden, "Internet Protocol, Version 6
               (IPv6) Specification", Internet Draft
               <draft-ietf-ipngwg-ipv6-spec-v2-01.txt>, November 1997.

   [RFC791]    Information Sciences Institute, "Internet Protocol",
               Internet RFC 791, September 1981.

   [RFC1122] RFC 1122, "Requirements for Internet hosts -
               communication layers".  R.T. Braden. Oct-01-1989.

   [RFC1349]   P. Almquist, "Type of Service in the Internet Protocol
               Suite", Internet RFC 1349, July 1992.

   [RFC1812]   F. Baker, editor, "Requirements for IP Version 4
               Routers", Internet RFC 1812, June 1995.

   [RFC2119]   S. Bradner, "Key words for use in RFCs to Indicate
               Requirement Levels", Internet RFC 2119, March 1997.

   [2BIT]      K. Nichols, V. Jacobson, and L. Zhang, "A Two-bit
               Differentiated Services Architecture for the Internet",
               ftp://ftp.ee.lbl.gov/papers/dsarch.pdf.


Appendix.  Examples of Class Selector Compliant PHB Implementations

   This appendix shows how three different mechanisms can be used to
   implement Class Selector Compliant PHBs.

A.1  Priority Queues

   This approach employs eight queues where each of the eight codepoints
   maps to a different queue.  Queues are ranked in priority order so
   that each queue, from the perspective of the next lower priority
   queue, implements a "low loss, low delay" forwarding behavior.  As an
   alternative, one can employ four queues, where bits 0 and 1 of the
   codepoint are used to select the queue.  Bit 2 of the codepoint is
   used as a drop preference within the queue.  RED is used within the
   queues according to its usual parameters, but with in-profile traffic
   having a higher min-threshold and max-threshold than out-of-profile
   traffic, and therefore experiencing a higher probability of timely
   delivery.  Out-of-profile traffic should consider the presence of
   lower order in-profile traffic in the calculation of drop
   probability [CLARK].

   The strength of this approach is that order is maintained within each
   precedence queue, but higher ordered traffic may be sent before lower
   ordered traffic.  It has a weakness, however, in that apart from
   admission control and policing, it affords lower precedence traffic
   no assurance of eventual transmission.


A.2  Round Robin Queuing

   Like the previous example, this approach employs a queue per
   codepoint, but each queue is emptied at some non-zero rate, in round-
   robin order, rather than being given simple priority service.

   The strength of this approach is that higher ordered traffic is, on
   average, queued for a shorter duration than lower ordered traffic.
   It also avoids the lockout issue that priority queuing systems
   experience.  A counter-intuitive scenario can occur, however, if a
   high rate queue is heavily utilized while a lower rate queue is
   under-utilized; a packet directed to the lower rate queue can
   actually be better protected from loss and variation in delay when
   placed in an empty or very short queue.


A.3  Virtual Circuit or Virtual Channel Selection

   The difference between this approach and Round Robin Queuing is
   somewhat academic.  If one has a serial line to a routing neighbor,
   and manages the link using a load sharing algorithm, the load sharing
   algorithm in some sense emulates the way the line would behave if it
   were in reality a number of different lines, or if it were one
   channelized line.  In a virtual circuit selection model, the
   emulation becomes reality - one deploys a set of rate-limited VCs to
   a routing neighbor, and uses them in the same way one would otherwise
   have used round-robin queues.

   The strengths and weaknesses are very similar to those of Round Robin
   Queuing, except that this allows one to capitalize on the
   capabilities of a link layer such as ATM or Frame Relay.


Authors' Addresses

   Kathleen Nichols
   Bay Networks Architecture Lab
   4401 Great America Parkway
   SC01-04
   Santa Clara, CA 95052-8185
   Phone:  +1-408-495-3252
   Fax:    +1-408-495-1299
   E-mail: knichols@baynetworks.com

   Steven Blake
   Torrent Networking Technologies (effective 7/15)
   E-mail: slblake@intercenter.net

   Fred Baker
   Cisco Systems
   519 Lado Drive
   Santa Barbara, California 93111
   Phone: (408) 526-4257
   Email: fred@cisco.com

   David L. Black
   The Open Group
   11 Cambridge Center
   Cambridge, MA 02142
   Phone: (617) 621-7347
   Email: d.black@opengroup.org