INTERNET-DRAFT                                        Mark W. Garrett,
ISSLL WG                                              Bellcore
Expires 26 September 1997
Expires: 25 January 1998
                                                      Marty Borden,
                                                      New Oak Communications

Interoperation of Controlled-Load Service and Guaranteed Services Service with ATM
                 <draft-ietf-issll-atm-mapping-02.txt>
                 <draft-ietf-issll-atm-mapping-03.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 "work in progress.'' progress."

   To learn the current status of any Internet-Draft, please check the
   ``1id-abstracts.txt''
   "1id-abstracts.txt" listing contained in the Internet- Drafts Shadow
   Directories on ftp.is.co.za (Africa), nic.nordu.net (Europe),
   munnari.oz.au (Pacific Rim), ds.internic.net (US East Coast), or
   ftp.isi.edu (US West Coast).

Abstract

   Service mappings are an important aspect of effective interoperation
   between Internet Integrated Services and ATM networks.

   This document provides guidelines for ATM virtual connection mapping service classes, and
   traffic management features and parameters to be used in support of the IP integrated services
   protocols.  The specifications include IP Guaranteed Service,
   Controlled-Load Service between Internet and ATM Forum UNI specification, versions
   3.0, 3.1 and 4.0.

   These
   technologies.  The service mappings are intended to facilitate useful for providing
   effective end-to-
   end interoperation and end-to-end Quality of Service for IP
   Integrated Services networks containing ATM subnetworks.
   We discuss the various features of

   The discussion and specifications given here support the IP
   integrated services protocols for Guaranteed Service (GS),
   Controlled-Load Service (CLS) and the ATM protocols, and
   identify solutions Forum UNI specification,
   versions 3.0, 3.1 and difficult issues 4.0.  Some discussion of compatibility and
   interoperation. IP best effort service
   over ATM is also included.

                              Table of Contents

   0.0 What's New in This Version .........................................   3

   1.0 Introduction .......................................................   x   3
       1.1 General System Architecture ....................................   x   5
       1.2 Related Documents ..............................................   x   7

   2.0 Discussion of ATM Major Protocol Features ................................   x for Traffic Management and QoS .............   8
       2.1 Service Category and Bearer Capability .........................   x   8
           2.1.1 Service Categories for Guaranteed Service ................   x   8
           2.1.2 Service Categories for Controlled Load ...................   x   9
           2.1.3 Service Categories for Best Effort .......................   x  10
       2.2 Cell Loss Priority Bit, Tagging and Conformance Definitions ....   x  12
       2.3 ATM Adaptation Layer ...........................................   x  13
       2.4 Broadband Low Layer Information ................................   x  13
       2.5 Traffic Descriptors ............................................   x  13
           2.5.1 Translating Traffic Descriptors for Guaranteed Service ...   x  14
           2.5.2 Translating Traffic Descriptors for Controlled Load Service  x 17
           2.5.3 Translating Traffic Descriptors for Best Effort Service .... x ....19
       2.6 QoS Classes and Parameters .....................................   x  20
       2.7 Additional Parameters -- Frame Discard Mode ....................   x  22

   3.0 Discussion of IP-IS Additional IP-Integrated Services Protocol Features ..............................   x ................  22
       3.1 Path Characterization Parameters for IP Integrated Services ....  22
       3.2 Handling of Excess Traffic .....................................   x
       3.2  23
       3.3 Use of AdSpec in Guaranteed Service with ATM ...................   x Adspec Parameters and Slack Term .....  24

   4.0 Discussion of Miscellaneous Items .................................. ................................................   x
       4.1 Units Conversion ...............................................   x

   5.0 Summary of ATM VC Setup Parameters for Guaranteed Service ..........   x
       5.1 Encoding GS Using Real-Time VBR ................................   x
       5.2 Encoding GS Using CBR ..........................................   x
       5.3 Encoding GS Using Non-Real-Time VBR ............................   x
       5.4 Encoding GS Using ABR ..........................................   x
       5.5 Encoding GS Using UBR ..........................................   x
       5.6 Encoding GS Using UNI 3.0 and UNI 3.1. .........................   x

   6.0 Summary of ATM VC Setup Parameters for Controlled Load Service .....   x
       6.1 Encoding CLS Using ABR .........................................   x
       6.2 Encoding CLS Using Non-Real-Time VBR ...........................   x
       6.3 Encoding CLS Using Real-Time VBR ...............................   x
       6.4 Encoding CLS Using CBR .........................................   x
       6.5 Encoding CLS Using UBR .........................................   x
       6.6 Encoding CLS Using UNI 3.0 and UNI 3.1. ........................   x

   7.0 Summary of ATM VC Setup Parameters for Best Effort Service .........   x
       7.1 Encoding Best Effort Service Using UBR .........................   x
       7.2 Encoding Best Effort Service Using Other ATM Service Categories    x

   8.0 Acknowledgements ...................................................   x

   Appendix 1  Abbreviations ..............................................   x
   REFERENCES
   References .............................................................   x
   AUTHORS' ADDRESSES
   Authors' Addresses .....................................................   x

0.0 What's New in This Version

   Corrections to VC setup parameter tables.

   Deleted specific QoS parameter values in tables.

   Section 3.1

New sections on path characterization parameters (Section 3.1), and
on handling of excess traffic. traffic (Section 3.2) have been added.  The
sections on translating traffic descriptors (Section 2.5) and QoS
paremeters (Section 2.6) have been substantially revised.  Minor
improvements were made in the mapping tables in Sections 5, 6, 7.

1.0 Introduction

   We consider the problem of providing IP Integrated Services [1] with
   an ATM subnetwork.  This document is intended to be consistent with
   the rsvp protocol [2] for IP-level resource reservation (although reservation, although it
   is, strictly speaking, independent of rsvp, since GS and CLS services
   can be supported through other mechanisms). mechanisms.  In the ATM network, we
   consider ATM Forum UNI Signaling, versions 3.0, 3.1 and 4.0 [3, 4,
   5].  The latter uses the more complete service model of The the ATM
   Forum's TM 4.0 specification [6, 7].

   This is a complex problem with many facets. problem.  In this document, we focus on the service
   types, parameters and signalling elements needed for service
   interoperation.  The resulting service mappings can be used to
   provide effective end-to-end Quality of Service (QoS) for IP traffic
   that traverses ATM networks.

   The IP services considered are Guaranteed Service (GS) [8] and
   Controlled Load Service (CLS) [9].  We also treat the default Best
   Effort Service (BE) in parallel with these.  Our recommendations for
   BE are intended to be consistent with RFC 1755 [10], and its revision
   (in progress) [11], which defines how ATM VCs can be used in support
   of normal BE IP service.  The ATM services we consider are:

       CBR           Constant Bit Rate
       rtVBR         Real-time Variable Bit Rate
       nrtVBR        Non-real-time Variable Bit Rate
       UBR           Unspecified Bit Rate
       ABR           Available Bit Rate
   (Note, Appendix 1 provides definitions for all abbreviations.)
   In the case of UNI 3.0 and 3.1 signaling, signalling, where these service are
   not all clearly distinguishable, we identify the appropriate
   available services.

   The service mappings which follow most naturally from the service
   definitions are as follows:

       Guaranteed Service    ->     CBR or rtVBR
       Controlled Load       ->     nrtVBR or ABR (with a minimum cell rate)
       Best Effort           ->     UBR or ABR

   For completeness we provide detailed mappings for all service
   combinations in Sections 5, 6, 7) and identify how each meets or
   fails to meet the requirements of the higher level IP services.  The
   reason for not restricting mappings to the most obvious or natural
   ones is that we cannot assume now that predict how widely available all of these
   services will always be ubiquitously
   available. as ATM deployment evolves.  A number of details,
   such as treatment of the difference in service definitions and the treatment of
   packets in excess of the flow traffic descriptor, make service
   mapping a relatively complicated subject, which cannot be expressed briefly and accurately
   at the same time. subject.

   The remainder of this introduction provides a general discussion of
   the system configuration and other assumptions.  Section 2 considers
   the relevant ATM protocol elements and their effects as related to
   Guaranteed, Controlled Load and Best Effort services (the latter
   being the default "service").  This section discusses features of the
   IP services as well; we chose to organize the main discussion by ATM
   features only because ATM is more complex in structure.  Section 3
   discusses a number of
   important remaining features of the IP services and how
   they can be handled on an ATM subnetwork.  Section 4 addresses a few miscellaneous problems
   which are an
   issue (units conversion) that is neither distinctly IP nor ATM.
   Section 5 gives the detailed VC setup parameters for Guaranteed
   Service, and considers the effect of using each of the ATM service
   categories.  Section 6 provides a similar treatment for Controlled
   Load Service.  Section 7 considers Best Effort service.

   This document is only a part of the total solution to providing the
   interworking of IP integrated services with ATM subnetworks.  The
   important issue of VC management, including flow aggregation, is
   considered in [12]. [12,18,19].  We do not consider how routing -- routing, QoS
   sensitive or not -- not, interacts with the use of VCs, especially in the case of
   multicast (or point-to-multipoint) flows. VCs.  We expect that a
   considerable degree of implementation latitude will exist, even
   within the guidelines presented here.  Many aspects of interworking
   between IP and ATM will depend on economic factors, and will not be
   subject to standardization.

1.1 General System Architecture

   We assume that the reader has a general working knowledge of IP, rsvp
   and ATM protocols.  The network architecture we consider is
   illustrated in Figure 1, below. 1.  An IP-attached host may send unicast
   datagrams to another host, or may use an IP multicast address to send
   packets to all of the hosts which have "joined" the multicast "tree".
   In either case, a destination host may then use RSVP to establish
   resource reservation in routers along the internet path for the data
   flow.

   An ATM network lies in the path (chosen by the IP routing), and
   consists of one or many more ATM switches.  It uses VCs SVCs to provide both
   resources and QoS within the ATM cloud.  These connections are set
   up, added to (in the case of multipoint trees), torn down, and
   controlled by the edge devices, which act as both IP routers and ATM
   interfaces, capable of initiating and managing VCs across the ATM
   user-to-network (UNI) interface.  The edge devices are assumed to be
   fully functional in both the IP int-serv/RSVP protocols and the ATM
   UNI protocols, as well as translating between them.

                                  ATM Cloud
                              ------------------
        H ----\              (                  )          /------- H
        H ---- R -- R -- E --( ATM Sw -- ATM Sw ) -- E -- R -- R -- H
        H ----/     |        (                  )          \
                    |         ------------------            \------ H
        H ----------R

              Figure 1:  Network Architecture with hosts Hosts (H),
                         Routers (R) and Edge Devices (E).

   When considering the edge devices with respect to traffic flowing
   from source to destination, the upstream edge device is called the
   "ingress" point and the downstream device the "egress" point.  The
   edge devices may be considered part of the IP internet or part of the
   ATM cloud, or both.  This is not an issue since they must provide
   capabilities of both environments.  The edge devices have normal RSVP
   capability to  They process RSVP messages, reserve resources,
   and maintain soft state (in the control path), and to classify and
   schedule packets (in the data path).  They also have the normal ATM
   capabilities to initiate ATM
   connections by signaling, signalling, and to accept or refuse connections signaled
   to them.  They police and schedule cells going into the ATM cloud.  An
   The service mapping function occurs when an IP-level reservation
   (RESV message) triggers the edge device to translate the RSVP service
   requirements into ATM VC (UNI) semantics.

   A range of VC management policies are possible, which determine
   whether a flow should initiate a new VC or join an existing one.  VCs
   are managed according to a combination of standards and local policy
   rules, which are specific to either the implementation (equipment) or
   the operator (network service provider).  Point-to-multipoint
   connections within the ATM cloud can be used to support general IP
   multicast flows.  In ATM, a point to multipoint connection can be
   controlled by the source (or root) node, or a leaf initiated join
   (LIJ) feature in ATM may be used.  Note, the topic of VC management
   and mapping of flows onto VCs is considered at length in another other issll
   working group draft [12]. drafts [12,18,19].

   Figure 2 shows the functions of an edge device, summarizing the work
   not part of IP or ATM abstractly as an InterWorking Function (IWF),
   and segregating the control and data planes.  (Note: for expositional
   convenience, policy control and other control functions are included
   as part of the admission control in the diagram.)

         IP                                                ATM
                               ____________________
                              |        IWF         |
                              |                    |
         admission       <--> | service mapping    | <-->  ATM
         control              | VC management      |       signalling &
                              | address resolution |       admission
                              |....................|       control
                              |                    |
         classification/      |ATM Adaptation Layer|       cell
         policing &      <--> | Segmentation and   | <-->  scheduling/
         scheduling           |  Reassembly        |       shaping
                              | Buffering          |
                               ____________________

                 Figure 2: Edge Device Functions showing the IWF

   In the logical view of Figure 2, some functions, such as scheduling,
   are shown separately, since these functions are required of both the
   IP and ATM sides.  However it may be possible in an integrated
   implementation to combine such functions.

   It is not possible to completely separate the service mapping and VC
   management functions.  Several illustrative examples come to mind:
   (i) Multiple integrated-services flows may be aggregated to use one
   point-to-multipoint VC.  In this case, we assume the IP flows are of
   the same service type and their parameters have been merged
   appropriately.  (ii) The VC management function may choose to
   allocate extra resources in anticipation of further reservations or
   based on an empiric of changing TSpecs.  (iii) There must exist a
   path for best effort flows and for sending the rsvp control messages.
   How this interacts with the establishment of VCs for QoS traffic may
   alter the characteristics required of those VCs.  See [12] [12,18,19] for
   further details on VC management.

   Therefore, in discussing the service-mapping problem, we will assume
   that the VC management function of the IWF can always express its
   result in terms of an IP-level service with some QoS and TSpec.  The
   service mapping algorithm, which is the subject of this document, algorithm can then identify the appropriate VC
   parameters, whether the resulting action is initiation of a uses new VC, the addition/deletion of a leaf to
   an existing multipoint tree, or the modification of an existing VC to
   one of another description. VCs.

1.2 Related Documents

   Earlier ATM Forum documents were called UNI 3.0 and UNI 3.1.  The 3.1
   release was used to correct errors and fix alignment with the ITU.
   Unfortunately UNI 3.0 and 3.1 are incompatible.  However this is in
   terms of actual codepoints, not semantics.  Therefore, descriptions
   of parameter values can generally be used for both.

   After 3.1, the ATM Forum decided began to release documents separately for
   each technical working group.  The Traffic Management and Signalling
   4.0 documents are available publically at ftp.atmforum.com/pub.  We
   refer to the combination of traffic management and signalling as
   TM/UNI 4.0, although specific references may be made to the TM 4.0
   specification or the UNI SIG Signalling 4.0 specification.

   Within the IETF area, IETF, related material includes the work of the rsvp [2],
   int-serv [1, 8, 9, 13, 14] and ion working groups [10, 11] of
   the IETF. 11].  Rsvp
   defines the resource reservation protocol (which is analogous to signaling
   signalling in ATM). Int-serv defines the behavior and semantics of
   particular services (analogous e.g., to the Traffic Management
   working group in the ATM Forum).  Ion defines interworking of IP and
   ATM for traditional Best Effort service, and covers all issues related to
   routing and addressing.

   A large number of ATM signaling signalling details are covered in RFC 1755
   [10], e.g., differences between UNI 3.0 and UNI 3.1, encapsulation, frame-
   relay
   frame-relay interworking, etc.  These considerations generally extend
   to IP over ATM with QoS as well.  Any description given in this
   document of IP Best Effort service (i.e. the default behavior) over
   ATM is intended to be consistent with RFC 1755 and it's extension for
   UNI 4.0 [11], and those documents are to be considered definitive.
   In some instances with non-best-effort services, certain IP/ATM
   features will diverge from the following RFC 1755.  The authors have
   attempted to note such differences explicitly.  (For example, best
   effort VCs are taken down on timeout by either edge device, while QoS
   VCs are only removed by the upstream edge device when the
   corresponding rsvp reservation is deleted.)

   RFC 1821 [15], represents an early discussions discussion of issues involved with
   interoperating IP and ATM protocols for integrated services and QoS.

2.0 Discussion of ATM Major Protocol Features for Traffic Management and QoS

   In this section, we discuss each of the items that must be specified
   in the setup of an ATM VC.  For each of these we discuss which
   specified items and values may be most appropriate for each of the
   three integrated services.

   The ATM Call Setup is sent by the ingress edge device to the ATM
   network to establish end-to-end (ATM) service.   This setup contains
   the following information.

       Service Category/Broadband Bearer Capability
       AAL Parameters
       Broadband Low Layer Information
       Calling and Called Party Addressing Information
       Traffic Descriptors
       QoS Class and/or Parameters
       Additional Parameters of TM/UNI 4.0

   We will discuss each of these, except addressing information, these as they relate to the translation of GS and
   CLS to ATM services.  Following
   the discussion  We do not discuss addressing at all, since it
   is (at least presently) independent of QoS.  Following the section on
   service categories, we discuss the tagging and conformance definitions
   for IP and ATM, ATM.  These do not appear explicitly as set-up parameters
   since the policing method used is implicit in the call setup.  We then continue with mappings of the
   other parameters and information elements.

2.1 Service Category and Bearer Capability

   The highest level of abstraction distinguishing features of ATM VCs
   is in the service category or bearer capability.  Service categories
   were introduced in TM/UNI 4.0; previously the bearer capability was
   used to discriminate at this level.

   In each version of the ATM specifications, these

   These parameters indicate the general properties required of a VC:
   whether there is a real-time delay constraint, whether the traffic is
   constant or variable rate, the applicable traffic and QoS description
   parameters and (implicitly) the complexity of some supporting switch
   mechanisms.

   For UNI 3.0 and UNI 3.1, there are only two distinct options for
   bearer capabilities (in our context):

       BCOB-A:  constant rate, timing required, unicast/multipoint;
       BCOB-C:  variable rate, timing not required, unicast/multipoint.

   There is a

   A third capability, BCOB-X, but in can be used as a substitute for the case of AAL5 (which
   we require -- see below) it can be used interchangeably above
   two capabilities, with its dependent parameters (traffic type and
   consistently
   timing requirement) set appropriately.  The distinction between the
   BCOB-X formulation and the "equivalent" (for our purposes) BCOB-A and
   BCOB-C constructs is whether the ATM network is to provide pure cell
   relay service or interwork with other technologies (with
   interoperable signalling), such as narrowband ISDN.  Where this
   distinction is applicable, the above two capabilities. appropriate code should be used (see
   [5] and related ITU specs, e.g., I.371).

   In TM/UNI 4.0 the service categories are:

       Constant Bit Rate (CBR)
       Real-time Variable Bit Rate (rtVBR)
       Non-real-time Variable Bit Rate (nrtVBR)
       Unspecified Bit Rate (UBR)
       Available Bit Rate (ABR)

   The first two of these are real-time services, so that rtVBR is new
   to TM/UNI 4.0.  The ABR service is also new to TM/UNI 4.0.  UBR
   exists in all specifications, except perhaps in name, through the
   ``best effort''
   "best effort" indication flag and/or the use of QoS Class 0.

   The encoding used Service Category in TM/UNI 4.0 is consistent with encoded into the earlier versions.
   For example, same signalled
   Information Element (IE) as the Bearer Capability in UNI 3.x, for the
   purpose of backward compatibilty.  Thus, we use the convention of
   referring to Service Category is indicated solely by (CBR, rtVBR, nrtVBR, UBR, ABR) for
   TM/UNI 4.0 (where the
   combination of bearer capability is implicit).  When we refer
   to the Bearer Capability and the Best Effort indication
   flag. explicitly (BCOB-A, BCOB-C, BCOB-X), we are
   describing a UNI 3.x signalling message.

   In principle, it is possible to support any foreseeable service through the use
   of BCOB-A/CBR.  This is because the CBR service is equivalent to
   having a ``pipe'' "pipe" with specified bandwidth/timing.  However, it may be
   desirable to make better use of the ATM network's resources by using
   other, less demanding, services when available.  (See RFC 1821 for a
   discussion of this [15].)

   2.1.1 Service Categories for Guaranteed Service
   There are two possible mappings for GS:

       CBR (BCOB-A)
       rtVBR

   GS requires real-time support, that is, timing is required. support.  Thus in UNI 3.x, the bearer class
   BCOB-A (or an equivalent BCOB-X formulation) must be used.  In TM/UNI
   4.0 either CBR or rtVBR is appropriate.  In both cases, GS would use a value of CLR
   appropriately low for the link (i.e., such that congestion losses are
   dominated by losses due to bit errors).  The use of rtVBR may
   encourage recovery of allocated bandwidth left unused by a source.
   It also accomm odates accommodates more bursty sources with a larger token bucket
   burst parameter, and permits the use of tagging for excess traffic
   (see Section 2.2).

   Neither the BCOB-C bearer class, nor nrtVBR, UBR, ABR are good
   matches for the GS service.  These provide no delay estimates and
   cannot guarantee consistently low delay for every packet.

   Specification of BCOB-A or CBR requires specification of a PCR.  The
   PCR should be specified as the the token bucket peak cell
   rate parameter, with
   appropriate conversion from bytes to cells (accounting for overhead),
   of the GS TSpec.  For both of these, (PCR).  In these cases, PCR is the network provides a nominal clearing rate of PCR with
   jitter toleration (bucket size) CDVT,
   specified in a network specific manner (see below). which is generally small.

   Specification of rtVBR requires the specification of two rates, SCR PCR and PCR. SCR.  This models
   bursty traffic with specified peak and average sustainable rates.  With rtVBR, it is appropriate to map the PCR to the line rate
   of incoming traffic and the SCR to the GS TSpec bucket rate.  The
   corresponding ATM token bucket sizes depth values are CDVT, in a network specific manner, and CDVT+BT,
   respectively for the PCR and SCR parameters (see below).
   respectively.

   2.1.2 Service Categories for Controlled Load

   There are three possible good mappings for CLS:

       CBR (BCOB-A)
       ABR
       nrtVBR (BCOB-C)

   Note that under UNI 3.x, only the first and third choices there are
   applicable. equivalent services to CBR and
   nrtVBR, but not ABR.  The first, with a CBR/BCOB-A connection,
   provides a higher level of QoS than is necessary, but it may be
   convenient to simply allocate a fixed-rate ``pipe'', "pipe", which should we expect to
   be ubiquitously supported in ATM networks.  However unless this is
   the only choice available, this will probably be wasteful of network
   resources.

   The nrtVBR/BCOB-C category is perhaps the best match, since it
   provides for allocation of bandwidth and buffers with an additional
   peak rate indication, similar to the CLS TSpec.

   The ABR category with a positive MCR aligns with the CLS idea of
   ``best
   "best effort with floor.'' a floor."  The ATM network agrees to forward cells
   with a rate of at least MCR, which should be directly converted from
   the token bucket rate of the receiver TSpec.  The bucket size
   parameter measures approximately the amount of buffer required at the
   IWF.

   The nrtVBR/BCOB-C category can also  This buffer serves to absorb the bursts allowed by the token
   bucket, since they cannot be used. passed directly into a ABR VC.

   The rtVBR category can be used, although the edge device must choose
   determine a value for CTD and CDV CDV.  Since there are no corresponding
   IP-level parameters, their values are set as a matter of local
   policy.

   The UBR category does not provide enough capability for Controlled
   Load.  The point of CLS is to allow an allocation of resources, which
   is facilitated by the token bucket traffic descriptor, and is
   unavailable in with UBR.

   2.1.3 Service Categories for Best Effort

   All of the service categories have the capability to carry Best
   Effort service, but the natural service category is UBR (or, in UNI
   3.x, BCOB-C or BCOB-X, with the best effort indication set).  A CBR
   or rtVBR clearly could be used, and since the service is not real-
   time, a nrtVBR connection could also be used.  In these cases the
   rate parameter used reflects a bandwidth allocation in support of the
   ingress edge device's best effort connectivity to the far egress edge
   router.  It would be normal for traffic from many source/destination
   pairs to be aggregated on this connection; indeed, since Best Effort
   is the default IP behavior, the individual flows are not necessarily
   identified or accounted for.  CBR may be a preferred solution in the
   case where best effort traffic is sufficiently highly aggregated that
   a simple fixed-rate pipe is efficient.  Both CBR and nrt-VBR provide
   bandwidth allocation which may be useful for billing purposes.

   An ABR connection could similarly be used to support Best Effort
   traffic.  The  Indeed, the support of data communications protocols such
   as TCP/IP is the explicit purpose for which ABR was specifically designed.  It is
   conceivable that a separate ABR connection would be made for
   different IP flows, although the normal case would probably have all
   IP Best Effort traffic with a common egress router sharing a single
   ABR connection.

   The rt-VBR service category may be considered less suitable, simply
   because both the real-time delay constraint and the use of SCR/BT add
   unnecessary complexity.

   See specifications from the IETF ion working group [10, 11] for
   related work on support of Best Effort service with ATM.

2.2 Cell Loss Priority Bit, Tagging and Conformance Definitions

   An

   Each ATM cell header carries the a Cell Loss Priority (CLP) bit.  Cells
   with CLP=1 are said to be ``tagged'' "tagged" or "marked" and have lower
   priority.  This tagging may be done by the source, to indicate
   relative priority within the VC, or by a switch, to indicate traffic
   in violation of policing parameters.  Options involving the use of
   tagging are decided at call setup time.

   A Conformance Definition is a rule that determines whether a cell is
   conforming to the traffic descriptor of the VC.  The conformance
   definition is given in terms of a Generic Cell Rate Algorithm (GCRA),
   also known as a "leaky bucket" algorithm, for CBR and VBR services.
   (UBR and ABR have network-specific network implementation-specific conformance
   definitions.  Note, the term "compliance" in ATM is used to describe
   the behavior of a
   connection.) connection, as opposed to "conformance", which
   applies to a single cell.)

   The network may tag cells which that are non-conforming, rather than
   dropping them only if the VC is set up to request set-up requests tagging and the network
   supports the tagging option.  When congestion occurs, a switch must
   attempt to discard tagged cells in preference to the discarding of CLP=0
   cells.  However, the mechanism for doing this is completely
   implementation specific.   Tagged  The behavior that best meets the
   requirements of IP Integrated Services is where tagged cells are
   treated with a
   behavior which is Best Effort as "best effort" in the sense that they are transported when
   bandwidth is available, queued when buffers are available, and
   dropped when the resources are overcommitted.  ATM standards, however, do
   not explicitly specify treatment of tagged traffic.  Providers of GS
   and CLS service with ATM subnetworks should ascertain the actual
   behavior of ATM implementation with respect to tagged cells.

   Since GS and CLS services require excess traffic to be treated as
   Best Effort,
   best effort, the tagging option should always be chosen (if
   supported) in the VC setup as a means of ``downgrading'' non-
   conformant cells. "downgrading" the cells
   comprising non-conformant packets.  However, the term ``best effort'' seems to "best effort"
   can be used
   with two distinguishable meanings interpreted in the int-serv specs. two distinct ways.  The first is that of as a service
   class that, in some typical scheduler implementations, would
   correspond to a separate queue.  Placing excess traffic in best
   effort in this sense would be giving it lower delay priority.  The
   other sense is more generic, meaning that the network would make a
   best effort to transport the traffic.  A reasonable expectation is
   that a network with no contending traffic would transport the packet,
   while a very congested network would drop the packet.  A packet that
   could be tagged with lower loss priority (such as with the ATM CLP
   bit) would be more likely to be dropped, but would not normally be transported out of order reordered
   with respect to the conforming portion of the flow.  Such a mechanism
   would agree with the latter definition of best effort, but not the
   former.

   In TM/UNI 4.0 tagging does not apply  This interpretation is left to the CBR or ABR services.
   However, there implementation.

   There are three conformance definitions of VBR service (for both
   rtVBR and nrtVBR) to consider.  In VBR, only the conformance
   definition VBR.3 supports tagging and applies the GCRA with rate PCR
   to the aggregate CLP=0+1 cells, and another GCRA with rate SCR to the
   CLP=0 cells.  Thus this  This conformance definition should always be used in
   support of with
   a VBR service supporting IP integrated services.  For UBR service,
   conformance definition UBR.2 supports the use of tagging, but a CLP=1
   cell does not imply non-conformance; rather, it may be a hint of used to
   indicate network congestion.

   Once an ATM connection is established, and

   In TM/UNI 4.0 tagging does not apply to the particular conformance
   definition is determined, the resulting policing action is mandatory.
   Since CBR or ABR services.
   More precisely, the conformance algorithm operates on cells, when mapping rates definitions listed in TM 4.0 for CBR
   and bucket sizes from IP services to corresponding ATM parameters, a
   correction needs to ABR do not use tagging.  Since conformance definitions are
   network specific, it may be made (at call setup time) for the ATM
   segmentation overhead.  Unfortunately this overhead, as a ratio,
   depends on packet length, possible that implementations of CBR or
   ABR with tagging can exist.  Wherever an ATM network does support
   tagging, in the overhead largest for small
   packets.  Thus the appropriate correction could be based sense of transporting CLP=1 cells on minimum
   packet size, expected packet size, or otherwise in a network specific
   manner, determined at the edge device IWF.  See Section 4.1. "best effort"
   basis, it is a useful and preferable mechanism for handling excess
   traffic.

   It is always better fo for the IWF to tag cells when it can anticipate
   that the ATM network would do so.  This is because the IWF knows the
   IP packet boundaries and can tag all of the cells corresponding to a
   packet.  If left to the ATM layer UPC, the network would inevitably
   carry
   drop some cells of packets which are worthless, because some other the cells from those of a packet are while carrying others, which would
   then be dropped due to non-conformance. by the receiver.  Therefore, the IWF, knowing the VC
   GCRA parameters, should always anticipate the cells which will be
   tagged by the ATM UPC and tag all of the cells uniformly across each
   affected packet.

2.3 ATM Adaptation Layer

   The AAL type 5 encoding must be used, as specified in RFC 1483 and
   RFC 1755. AAL5 requires specification of the maximum SDU size in both
   the forward and reverse directions. Both GS and CLS specify a maximum
   packet size as part of the TSpec and this value shall be used as the
   maximum SDU in each direction for unicast connections, but only in
   one direction and for
   unidirectional point-to-multipoint connections, which are
   unidirectional. connections.  When more than one flow multiple flows
   are aggregated into a single VC, the M parameters of the receiver
   TSpecs are merged according to yield the largest packet size.  In no case
   can this exceed 65535 (or, of course, the MTU of rules given in the link). GS and CLS specs.

2.4 Broadband Low Layer Information

   The B-LLI Information Element is transferred transparently by the ATM
   network between the edge devices and is used to specify the
   encapsulation method.  Multiple B-LLI IEs may be sent as part of
   negotiation.  The default encapsulation LLC/SNAP [16] must be
   supported as specified in RFC 1577 [17] and RFC 1755.  Additional
   encapsulations are discussed in 1755 [10].  See RFC
   1755 and we refer to the
   discussion there. for information on additional encapsulations.

2.5 Traffic Descriptors

   The ATM traffic descriptor always contains specification of a peak cell rate (PCR) (in
   (for each direction).  For variable rate services it also contains specification of a
   sustainable cell rate (SCR) and maximum burst size (MBS).  The SCR
   and MBS form a leaky bucket pair (rate, depth), while the bucket
   depth parameter for PCR is CDVT.  Note that CDVT is not signaled
   explicitly, but is determined by the network operator, and serves as
   a measure of the jitter imposed by the network.

   Since CDVT is not signaled, and is generally presumed to be small, the leaky
   bucket traffic descriptor (TSpec) small (equivalent to a few
   cells of the Internet service token bucket depth), and cannot
   always be directly mapped into PCR/CDVT parameters. set independently for
   each connection, it cannot be used to account for the burstiness
   permitted by b of the IP-layer TSpec.  Additional buffering is needed
   at the IWF to account for the depth of the token bucket.

   The ATM Burst Tolerance (BT) is related equivalent to MBS (see TM 4.0 [6] for details).
   Roughly, they
   the exact equation).  They are both expressions of the bucket depth
   parameter that goes with SCR.  The units of BT is time while the
   units of MBS is cells.  Since both SCR and MBS are signalled, they
   can be computed directly from the IP layer traffic description.  The
   specific manner in which resources are allocated from the traffic
   description is implementation specific.  Note that when translating
   the traffic parameters, the segmentation overhead and minimum policed
   unit need to be taken into account (see Section 4.2 4.1 below).

   In ATM UNI SIG Signalling 4.0 there are the notions of Alternative
   Traffic Descriptors and Minimal Traffic Descriptors.  Alternative
   Traffic Descriptors enumerate other acceptable choices for traffic
   descriptors and are not considered here.  Minimal Traffic Descriptors
   are used in ``negotiation,'' "negotiation," which refers to the specific way in which
   an ATM connection is set up.  Very roughly it works like this,  To illustrate, roughly, taking PCR as
   an example: A minimal PCR and a requested PCR are signalled, the
   requested PCR being the usual item signalled, and the minimal PCR
   being the absolute minimum that the source edge device will accept.
   When sensing the existence of both minimal and requested parameters,
   the intermediate switches along the path may reduce the requested PCR
   to a ``comfortable'' "comfortable" level.  This choice is part of admission control,
   and is therefore implementation dependent.  If at any point the
   requested PCR falls below the minimal PCR then the call is cleared.
   Minimal Traffic Descriptors can be used to present an acceptable
   range for parameters and ensure a higher likelihood of call
   admission.  Whether anything more specific about
   Minimal Traffic Descriptors needs to be said here is left for further
   study (FFS).  In general, our discussion of connection parameters
   assumes the values resulting from successful connection setup.

   The Best Effort indicator (used only with UBR) and Tagging indicators
   are also part of the signaled information element (IE) containing the
   traffic descriptor.  In the UNI SIG 4.0 traffic descriptor IE there is an
   additional parameter, the Frame Discard indicator (see indicator, which is discussed
   below in Section
   2.7). 2.7.

   2.5.1 Translating Traffic Descriptors for Guaranteed Service

   For Guaranteed Service there is a peak rate, p, a the source Tspec TSpec contains peak rate, rate and
   and bucket depth parameters, p_s, r_s, a b_s.  The receiver Tspec rate TSpec
   contains corresponding parameters p_r, r_r, and an b_r.  The (receiver)
   Rspec also has a rate, R.  The two Tspec different TSpec rates are intended
   to support receiver heterogeneity, in the sense that different receivers can
   accept different rates representing different subsets of the sender's
   traffic.  In this document we leave  Whenever rates from different receivers differ, the values
   will always be merged appropriately before being mapping into ATM
   parameters.

   Note that when the sender and receiver TSpec rates r_s, r_r differ,
   there is no mechanism specified (in either rsvp or the int-serv
   specs) for indicating which subset of the traffic is to be
   transported.  Implementation of this feature for further study (FFS), is therefore completely
   network specific.  Hence the ambiguity in how policing and assume scheduling
   use the two Tspec rates are
   always identical. is an inherent and currently unresolved issue in
   IP-IS technology.

   The Tspec receiver TSpec rate describes the traffic itself, for which resources are
   to be reserved, and
   is may be used for policing, while the Rspec rate
   (which cannot be smaller) is the allocated service rate. bandwidth (or
   strictly speaking, a lower bound on this).  A receiver increases R
   over r r_r to reduce the delay.

   When mapping Guaranteed Service onto a rtVBR VC, the ATM traffic
   descriptor parameters (PCR, SCR, MBS) can often be set within the following
   bounds:

                  R <= cannonically
   as:

                                PCR <= min(p, line rate)
                  r <= = p_r
                                SCR <= PCR
                  0 <= = R
                                MBS <- b.

   Note that = b_r.

   There are a receiver can choose R > p number of conditions that may lead to lower the delay.  This
   leaves the first equation somewhat subject different choices.
   The following discussion is not intended so much to interpretation.  If a
   receiver chooses R > line rate, it seems clear that the admission
   control would simply reject set hard
   requirements, but to provide some interpretation and guidance on the reservation.
   bounds of possible parameter mappings.  The ingress edge device has
   generally includes a buffer preceding preceeding the ATM network which must interface.
   This buffer can be
   sufficient used to absorb bursts arriving faster than they can be admitted
   into that fall within the IP-
   level TSpec, but not within the ATM network.  For example, parameters may be set as PCR = R,
   SCR = r, MBS = b. traffic descriptor.  The edge device buffer of size b would absorb a
   burst sent at any IP-level peak rate.  Although this buffer exists,
   the ATM network must accept bursts at rate PCR, at least R, to ensure minimal
   requirement for guaranteed service is that the edge device delay is no greater than b/R.  Since in this buffer
   is
   may not in exceed b/R, and the delays within the ATM network, its delay is not included network must be
   accurately accounted for in D_ATM.

   For GS over CBR, the service rate is mapped to the PCR parameter,
   using values of Adspec parameters C and D
   advertised by the same constraint for PCR given above.  The ingress router (see Section 3.3 below).

   In general, if either an edge device again
   requires adequate buffering to accommodate the TSpec bucket depth and
   ensure delay before entering buffer of size b_r exists or the
   ATM network of no more than b/R.  If
   PCR maximum burst size (MBS) parameter is greater than R, at least b_r, then the buffer requirement may be relaxed
   accordingly.

   2.5.2 Translating Traffic Descriptors for Controlled Load Service

   Controlled Load service has a peak rate, p, a Tspec rate, r, and a
   corresponding bucket depth parameter, b.  The ATM traffic
   various rate parameters
   for nrtVBR service category are constrained by

                  r will generally exhibit the following
   relationship:

                     r_r <= SCR <= R <= PCR <= min(p, line rate)
                  0 APB <= MBS <- b.

   For ABR VCs, the Tspec line rate would be used

                     r_r <=       p_r       <= APB

   APB refers to set the minimum cell General Characterization Parameter,
   AVAILABLE_PATH_BANDWIDTH, which is negotiated in the Adspec portion
   of the PATH message.  APB reflects the narrowest bottleneck rate (MCR) parameter.  The bucket depth parameter does not map
   directly to a signalled ATM parameter,
   along the path, and so is always bounded by the edge device local line rate.  The
   receiver must have choose a
   buffer of at least b bytes.

   For CBR, the Tspec peak rate sets a lower bound on PCR, and again, the
   available buffering in the edge device must be adequate to
   accommodate possible bursts.

   2.5.3 Translating Traffic Descriptors for Best Effort Service

   For Best Effort service, there is no traffic description.  The UBR
   service category allows negotiation of PCR, simply to allow greater than APB for the
   source
   reservation to discover the smallest physical bottleneck along the path.

2.6 QoS Classes and Parameters

   In TM/UNI 4.0 the three QoS parameters may be individually signalled.
   These parameters are accepted, although the Cell Loss Ratio (CLR), Cell Transfer Delay
   (CTD), and Cell Delay Variation (CDV).  In UNI 3.x source peak rate, p_s, could
   be higher, as the setup message
   includes only source does not know the QoS Class, which value of APB.  There is essentially an index no
   advantage to a
   network specific table allocating any rate above APB of values course, so it is an
   upper bound for these three all the other parameters.  A
   network provider may choose

   We might normally expect to associate other parameters, such find R <= p_r, as
   Severely Errored Cell Block Ratio, but these are less well understood
   and accepted compared would be necessary for
   the simple mapping of PCR = p_r, SCR = R given above.  However, a
   receiver is free to choose R > p_r to lower the basic loss, GS delay and jitter parameters
   mentioned here.  The ITU has recently included a standard [8].  In
   this case, PCR cannot be set of
   parameter values for below R, because a (small) number burst of QoS classes in size b
   arriving into the latest
   version buffer must be cleared at rate R to keep the first
   component of Recommendation I.356, October 1996.  The network provider GS delay down to b/R.  So here we will have PCR = R.

   In the case R <= p_r, we may still choose R <= PCR < p_r.  The edge
   device buffer is then necessary (and sufficient) to define further network-specific QoS classes in addition absorb the bursts
   (limited to these. size b_r) which arrive faster than they depart.  For
   example, it may be the case that the cost of the ATM VC depends on
   PCR, while the cost of the Internet service reservation is not
   strongly dependent on the IP-level peak rate.  The problem user may the have
   an incentive to set p_r to APB, while the operator of agreement between network providers the IP/ATM edge
   router has an incentive to reduce PCR as much as possible.  This may
   be a realistic concern, since the charging models of IP and ATM are
   historically different as far as usage sensitivity, and the value of
   p_r, if set close to APB, could be many times the definition nominal GS
   allocated rate of QoS classes R.  Thus, we can set PCR to R, with a buffer of
   size b, with no loss of traffic, and no violation of the GS delay
   bound.

   A more subtle, and perhaps controversial case is completely unaddressed where we set SCR to date.  We
   a value below R.  The major feature of the GS service is to allow a
   receiver to specify the allocated rate R to be larger than the rate
   r_r sufficient to transport the traffic, in order to lower the
   queueing delay (roughly) from b/r + C_TOT/r + D_TOT to b/R + C_TOT/R
   + D_TOT.  To effectively allocate bandwidth R to the flow, we set SCR
   to match R.  (Note it is unnecessary in any case to set SCR above R,
   so the relation, SCR <= R, is still true.)  It is possible to set SCR
   to a value as low as r_r, without violating the delay bounds or
   overflowing the edge device buffer.  With PCR = R, a burst of size b
   will be buffered and sent into the ATM network at rate R, so the last
   byte suffers delay only b/R.  Any further traffic will be limited to
   rate r_r, which is SCR, so with the arriving and departing rates
   matched, its delay will also be no more than b/R.

   While this scenerio meets the GS service requirements, the penalty
   for allocating SCR = r_r rather than R is that the delay in the ATM
   network will have a component of MBS/SCR, which will be b/r rather
   than b/R, contained in the D term advertised for the ATM sub-network
   (see further discussion in Section 3.3 below).  It is also true that
   allocating r instead of R in a portion of the path is rather against
   the spirit of GS.  As mentioned above, this mapping may however be
   useful in practice in the case where pricing in the ATM network leads
   to different incentives in the tradeoff between delay and bandwidth
   than those of the user who buys IP integrated services.

   Another point of view on parameter mapping suggests that SCR should
   merely reflect the traffic description, hence SCR = r_r, while the
   service requirement is expressed in the QoS parameter as CDV = b/R.
   Thus the ATM network may internally allocate bandwidth R, but it is
   free to use other methods as well to achieve the delay constraint.
   Mechanisms such as statistical flow/connection aggregation may be
   implemented in the ATM network and hidden from the user (or parameter
   mapping module in the edge router) which sees only the interface
   implemented in the signaled parameters.

   Note that this discussion considers an edge device buffer size of
   b_r.  In practice, it may be necessary for the AAL/segmentation
   module to buffer M bytes in converting packets to cells.  Also an
   additional amount of buffer equal to C_sum + R D_sum is generally
   necessary to absorb jitter imposed by the upstream network [8].

   With ATM, it is possible to have little or no buffer in the edge
   router, because the ATM VC can be set to accept bursts at peak rate.
   This may be unusual, since the edge router normally has enough buffer
   to absorb bursts according to the TSpec token bucket parameters.  We
   consider two cases.  First, if PCR >= p_r, then MBS can be set to b_r
   and no buffering is necessary to absorb normal bursts.  The extra
   buffering needed to absorb jitter can also be transferred to MBS.
   This effectively moves the buffering across the UNI into the ATM
   network.

   For completeness, we consider an edge router with no burst-absorbing
   buffers and an MBS parameter of approximately zero.  In this case it
   is sufficient to set the rate parameters to PCR = SCR = max (R, p_r).
   This amounts to peak-rate allocation of bandwidth, which will not
   usually be very cost effective.  One reason for mentioning this case
   might be that IP routers and ATM switches differ so substantially in
   their buffering designs that IP-level users typically specify much
   larger burst parameters than can be handled in the ATM subnet.
   Peak-rate bandwidth allocation provides a means to work around this
   problem.  It is also true that intermediate tradeoffs can be
   formulated, where the burst-absorbing buffer is less than b bytes,
   and SCR is set above R and below p_r.  Note that jitter-absorbing
   buffers (C_sum + R D_sum) can not be avoided, generally, by
   increasing ATM rates, unless SCR is set to exceed the physical line
   rate(s) into the edge device for the flow.

   For GS over CBR, the value of PCR may be mapped to the Rspec rate R,
   if the edge device has a buffer of size b_r.  With little or no burst
   buffering, the requirements resemble the zero-buffer case above, and
   we have PCR = max (R, p_r).  Additional buffers sufficient to absorb
   network jitter, given by C_sum, D_sum, must always be provided in the
   edge router, or in the ATM network via MBS.

   2.5.2 Translating Traffic Descriptors for Controlled Load Service

   The Controlled Load service TSpec has a peak rate, p, a "token
   bucket" rate, r, and a corresponding token bucket depth parameter, b.
   The receiver TSpec values are used to determine resource allocation,
   and a simple mapping for the nrtVBR service category is given by,

                                PCR = p_r
                                SCR = r_r
                                MBS = b_r.

   The discussions in the preceeding section on using edge device
   buffers to reduce PCR, increasing buffers to reduce PCR and trading
   off between such buffers and MBS, apply generally to the CLS over
   nrtVBR case as well.  Extra buffers to accommodate jitter accumulated
   (beyond the b_r burst size allowed at the source) must be provided.
   For CLS, there are no Adspec parameters C and D, so the estimation of
   such buffers is an implementation design issue.

   For ABR VCs, the TSpec rate r_r is used to set the minimum cell rate
   (MCR) parameter.  Since there is no corresponding signalled bucket
   depth parameter, the edge device must have a buffer of at least b_r
   bytes.  Since the actual transfer rate can vary substantially with
   ABR, the buffering should not be made so large that the, in an
   attempt to avoid loss, that delays exceed higher-layer timeouts,
   e.g., TCP retransmission.

   For CBR, the TSpec rate r_r sets a lower bound on PCR, and again, the
   available buffering in the edge device must be adequate to
   accommodate possible bursts of b_r.

   The requirement for CLS that network delays approximate "best-effort
   service under unloaded conditions", is interpreted here to mean that
   an allocation of (at least) r_r, resulting in the last byte of a
   burst of size b_r having delay approximately b_r/r_r, is sufficient.
   A network element e.g., with no cross-traffic, work conserving
   scheduling and output link rate of r_L might provide delays in the
   range from M/r_L to b_r/r_L, which may be much better.

   2.5.3 Translating Traffic Descriptors for Best Effort Service

   For Best Effort service, there is no traffic description.  The UBR
   service category allows negotiation of PCR, simply to allow the
   source to discover the smallest physical bottleneck along the path.
   (The ingress edge router should set PCR to the ATM line rate, and may
   wish to make use of the returned value when the VC is set up.)  Often
   a service provider will want to statically configure large VCs with a
   certain bandwidth allocation to handle all best effort traffic
   between two edge routers.  ABR, CBR or nrtVBR VCs are appropriate for
   this with traffic parameters set to comfortably accommodate the
   expected traffic load.  See [10,11].

2.6 QoS Classes and Parameters

   In UNI 3.x the quality of service is indicated by a single parameter
   called "QoS Class," which is essentially an index to a network
   specific table of values for the actual QoS parameters.  In TM/UNI
   4.0 three QoS parameters may be individually signalled, and the
   signalled values override those implied by the QoS Class, which is
   still present.  These parameters are the Cell Loss Ratio (CLR), Cell
   Transfer Delay (CTD), and Cell Delay Variation (CDV) [6].

   A network provider may choose to associate other parameters, such as
   Severely Errored Cell Block Ratio, with a QoS Class definition, but
   these cannot be signalled individually.  The ATM Forum UNI 3.0, 3.1
   and TM 4.0 specs, following prior ITU specs, give vague qualitative
   definitions for QoS Classes 1 to 4.  (QoS Class 0 is well-defined as
   "no QoS parameters defined".)  Since our mapping is based on these
   specifications, we generally follow this guidance by setting QoS
   Class value to 0 for UBR and ABR (as required), 1 for CBR and rtVBR
   and 3 for nrtVBR.  Note that the QoS Class follows the ATM service
   category, and not the IP service, to avoid combination that are
   unlikely to be supported.  For example, if only nrtVBR is available
   for GS, then choosing QoS Class = 1 would probably result in
   connection failure, rather than a way to add real-time behavior to an
   inherently non-real-time service.

   The ITU has recently included a standard set of parameter values for
   a (small) number of QoS Classes in the latest version of
   Recommendation I.356, October 1996.  Network providers may choose to
   define further network-specific QoS Classes in addition to these.
   Note that the QoS class definitions in the new I.356 version may not
   align with the model we follow from the UNI specs.  Apart from these
   definitions, the problem of agreement between network providers as to
   the definition of QoS Classes has not, to our knowledge, been
   addressed.

   The ATM QoS parameters have no explicitly signalled IP layer
   counterparts.  The values that should be signalled in the ATM network
   are determined by knowledge of certain network characteristics and
   the IP service definitions.

   The ingress edge router must keep a table of QoS information for the
   set of egress routers that it may establish VCs with.  This table may
   be simplified by using default values, but it will probably be good
   network practice to maintain a table of current data for the most
   popular egress points.  An ATM network generally needs to have some
   way to propose initial value for CDV and CTD, even if changed by
   negotiation; so by positing such a table, we are not creating any new
   design burden.  Cached information can be updated when VCs are
   successfully established, and to the extent that IP-layer
   reservations can wait for VCs to complete, the values can be refined
   through iterated negotiation.  In general the construction of this
   table is implementation specific.

   Both GS and CLS require that losses of packets due to congestion
   should be minimized, so that the loss rate is approximately the same
   as for an unloaded network.  The characteristic loss behavior of the
   link-layer medium not due to congestion (e.g., bit errors or fading
   on wireless channels) determines the order of the permitted packet
   loss rate.  The ingress edge device will choose a value of CLR that
   provides the appropriate IP-level packet loss rate.  The CLR value
   may be uniform over all egress points in the ATM network, or may
   differ, e.g., when wireless or satellite ATM links in the path.  The
   determination of CLR should account for the effects of packet size
   distribution and ATM Frame Discard mode (which can change the
   effective packet loss rate by orders of magnitude, given the same
   underlying cell loss rate [20]).

   The ingress router will also tabulate values for the Minimum Path
   Latency (MPL) and estimated queueing delays (D_ATM) for each egress
   point.  The latter will be used as part of the Adspec "D" parameter
   for GS, but its use here applies to CLS as well.  MPL represents all
   non-congestion related delays, including propagation delay.  D_ATM
   accounts for the variable component of delays in the ATM network.
   (It may depend on parameters such as CDVT, etc.) Hence, when a VC is
   set up, the delay-related QoS parameters are given by

                                CDV = D_ATM
                                CTD = D_ATM + MPL.

   (CDV and CTD may be increased by the slack term in GS, see Section
   3.3 below.)  For rtVBR, the value of CDV will generally have a
   component of MBS/SCR analogous to the b/R term in the delay of GS
   service.  It may have other components that depend on the ATM switch
   implementation.  In cases where the ATM network uses statistical
   resouce allocation methods, it may be possible to establish VCs with
   CDV less than MBS/SCR.  This capability should be reflected in the
   D_ATM values advertised in GS and used to determine CDV in for VCs
   supporting both GS and CLS.

   It is interesting (and perhaps unfortunate) to note that in a typical
   GS/rtVBR service, the delay bound advertised can contain two
   components of b/R instead of one.  Consider the case where SCR = R
   and MBS = b.  Parekh's theory, which is the basis of the GS delay
   formula [8] states that the b/R delay term occurs only once, because
   once a burst of size b has been served by a congested node at rate R,
   the packets will not arrive at a subsequent node as a single burst.
   However, we can't tell if this bottleneck will occur in the ATM
   network or elsewhere in the IP network, so the declaration of CDV
   must account for it.  Once CDV is set, the ATM network can impose
   that delay.  Since the delay b/R can also occur elsewhere, it cannot
   be removed from the first term of the GS delay formula.  The ATM b/R
   delay component appears in the third term, D_tot.  See Section 3.3
   below for more on GS Adspec parameters.  This effect may be
   unapparent when the ATM network employs more efficient statistical
   resource allocation schemes.

2.7 Additional Parameters -- Frame Discard Mode

   TM/UNI 4.0 allows the user to choose a mode where the ATM network is
   aware, for the purpose of congestion management, of PDUs larger than
   an ATM cell (i.e., AAL PDUs that correspond in our context to IP
   packets).  This facilitates implementation of algorithms such as
   partial packet discard, where a dropped cell causes subsequent cells
   in the AAL5 PDU to be dropped as well.  Several other applicable
   buffer management schemes have been proposed [20, 21].

   Frame discard can improve efficiency and the performance of end-to-
   end protocols such as TCP, since the remaining cells of a damaged PDU
   are generally useless to the receiver.  For IP over ATM, Frame
   Discard should always be indicated, if available.

3.0 Additional IP-Integrated Services Protocol Features

3.1 Path Characterization Parameters for IP Integrated Services with ATM

   This section discusses the setting of General Characterization
   Parameters (GCPs) at an ATM egress edge router.  GCPs are signalled
   from source to destination, and modified by intermediate nodes using
   the Adspec portion of PATH messages in rsvp.  The GS-specific Adspec
   parameters are discussed below in Section 3.3.  These parameters are
   denoted as <x,y> where x is the service and y is the parameter
   number.  Service number 1 indicates default or general parameter
   values.  Please refer to [22] for definitions and details.

   The IS break bit <1,2> should, of course, be left alone by
   implementations following these guidelines (as they are presumably
   IS-aware).  Similarly, the router should always increment IS_HOPS
   <1,4>.  The GS and CLS service-specific break bits, <2,2> and <5,2>
   respectively, should be set if the support of the service is
   inadequate.  In general GS is adequately supported by CBR (BCOB-A)
   and rtVBR service categories, and not adequately supported by UBR,
   ABR and nrtVBR because delays are not controlled.  CLS may be
   adequately supported by all service categories except UBR (or Best
   Effort in UNI 3.x).  See Sections 5, 6 for further discussion.

   For GS, the ATM network must meet the delay performance advertised
   through the Adspec parameters, MPL, C, and D.  If it cannot
   predictably meet these requirements, the GS break bit should be set.
   Similarly both break bits should be set if reservations are honored,
   but sufficient resources to avoid congestion loss are not allocated
   in practice.  If the service break bits are not set, then the
   corresponding service hop counters, <2,4>, <5,4>, should be
   incremented.

   The Available Path Bandwidth (APB) parameters <x,6> indicate the
   minimum physical bottleneck rate along the path.  This may be
   discoverable in an ATM network as the negotiated PCR value for a UBR
   VC along the path.  This value should be corrected for AAL, ATM and
   physical-layer headers, as necessary, to reflect the effective IP
   datagram bandwidth.  With ATM, it is possible that there is some
   policy limitation on the value of PCR, below the physical link
   bottleneck.  In this case, the advertised value of APB (in general
   and for each service if different) should reflect this limit, since
   excess traffic beyond this rate will adopt a convention expressed in UNI 3.x, that assumes be dropped.  (Note that QoS
   class 1 is appropriate for low-delay, low-loss CBR connections, and
   QoS class 3 there is appropriate
   no tagging of traffic in excess of PCR for variable rate connections with loss
   and delay roughly appropriate TM/UNI 4.0.)  These values
   should generally be cached by the ingress router for non-real-time data applications.
   Note the set of
   egress routers that it typically needs to establish VCs to.  The
   Adspec parameters <x,6> are only adjusted down, to reflect the QoS class definitions in
   minimum as the new I.356 version may not
   align with composed value.

   In the case of a multipoint VC, the value of several parameters can
   be different for each egress point.  In this model.

   Since no IP layer counterparts to case, the IWF at the
   egress routers must correct these ATM QoS parameters exist values in
   any of the IP services, PATH messages as they
   exit the ATM network.  This is the only case where the egress router
   needs to operate on rsvp control messages.  (A similar correction
   must be set by policy of the edge
   device. implemented for any non-rsvp set-up mechanism).  The QoS classes can be chosen relatively easily.  QoS class
   1 should be used with Guaranteed Service and QoS class 3 should be
   used with Controlled Load Service.  Best Effort Service always gets
   QoS class 0, which is unspecified QoS by definition.  There
   parameters that require such correction are two
   issues which amount to specifically the same thing: First,
   Available Path Bandwidth (APB), the choice of
   individually signalled parameter values (under TM/UNI 4.0) Minimum Path Latency (MPL), the
   Path MTU (although for GS ATM/AAL5 this may typically be constant), and
   CLS is
   the edge device policy. ATM-specific components of the GS Adspec parameters C_ATM and
   D_ATM.

   The second issue is choosing
   parameter ingress router table must store values for the two QoS classes, which is ATM-network MPL
   <x,7> for the various egress points.  The composed values <x,8> are
   formed by addition and forwarded along the path.  In the cases where
   ATM network
   policy. routing chooses different paths for VCs to a given egress point,
   depending on the service category, the table will generally reflect
   different values for each service.  If the same ATM network operator controls both, then these
   problems are identical; if not, an agreement is very large
   and complex, it may become difficult to make predict the values
   identical would be extremely desirable.

   Note routes that we have mapped QoS class 1 and 3 onto Guaranteed and
   Controlled Load service respectively. VCs
   will take once they are set up.  This is regardless of what
   service category is used.  So when running CLS over a CBR pipe, it
   would not could be inappropriate to use QoS class 3.  This leaves the a significant source
   of misconfiguration, resulting in discrepencies between GS delay
   unspecified (or much looser than with QoS 1).  These comments should
   be taken as preliminary, as these issues are far from clear,
   advertisements and
   industry consensus should actual results.  The RSpec Slack term may be sought.

2.7 Additional Parameters -- Frame Discard Mode

   In TM/UNI 4.0 ATM allows the user to choose a mode where a dropped
   cell causes all cells up to the last remaining
   useful in the AAL5 PDU mitigating this problem.

   AAL 5 will support any message size up to be
   also dropped.  This improves efficiency and the behavior of end-to-
   end protocols such as TCP, since 65,535 bytes, so setting
   the remaining cells of a damaged PDU
   are useless AAL SDU to the receiver.  For IP over ATM, Frame Discard receiver TSpec M parameter value should
   always generally
   not be used in both directions, if available, for all services.

3.0 Discussion of IP-IS Protocol Features

3.1 Handling of Excess Traffic

   (Placeholder a issue.  In the PATH Adspec, however, the PATH_MTU parameter
   <x,10> for text.)

           Reiterate that whole packets each service should be tagged, See Section
   2.2. set to 9180 bytes, which is the
   default MTU for AAL 5.

3.2 Use Handling of AdSpec Excess Traffic

   CLS requires and GS recommends that network elements transport
   traffic in Guaranteed Service with ATM

   The AdSpec is a feature excess of Guaranteed Service which allows the TSpec parameters whenever physical resources
   (bandwidth, buffers and processing) are available.  While excess
   traffic should be supported on a receiver
   to calculate best effort basis, it should not
   interfere with the worst-case delay associated QoS (delay and loss) of conforming CLS and GS
   traffic, nor with normal service of non-reserved best effort traffic.

   There are several solutions with ATM: the most attractive is to use a GS flow.  Three
   quantities, C, D,
   VBR service category (with an appropriate conformance definition) and MPL, are accumulated (by simple addition
   tag excess traffic as low priority using the CLP bit.  This avoids
   reordering of
   components, one for each network element) the flow, but requires care in the PATH message from
   source to receiver.  The resulting values can design of the egress
   router scheduler.  To avoid reordering, the excess traffic would be different for each
   unique receiver.  The maximum delay
   queued with confoming traffic.  A threshold must be used to ensure
   that conforming traffic is then found not unnecessarily delayed by

           delay <=  b/R + C/R + D + MPL

   The Maximum Path Latency (MPL) includes propagation delay and any
   other unavoidable system delays.  (We neglect the effect of maximum
   packet size and peak rate here; see the GS specification [8] excess.
   Also, for GS, the
   more detailed equation.)  The service rate requested by the receiver,
   R, can extra delay that would be greater than the sender's Tspec rate, r.  The effect of incurred due to excess
   traffic below the
   larger R is threshold would have to allocate more bandwidth and, through this equation,
   lower be accurately reflected in
   the packet delay.  The burst size, b, is delay advertisement.  Note that the egress router should
   uniformly tag all the leaky bucket
   parameter from cells of each non-conforming packet, rather
   than letting the Tspec, and ATM network apply tagging due to ATM-level non-
   conformance.

   There is not changed no requirement in ATM standards that tagged cells, marked
   either by the receiver in user or by policing, must be transported if possible.
   Therefore, the
   Rspec.

   The values operator of C and D which a an edge router advertise will depend on both supporting IP-IS should
   ascertain the particular packet scheduling algorithm used actual behavior of the ATM equipment in the router, and path, which
   may span multiple administrative domains in the characteristics ATM network.  If
   tagged cells are simply dropped at some point, regardless of load,
   then the subnet attached to the router.  We assume
   here that each router (or operator may consider setting the source host) takes responsibility break bit, at least for
   its downstream subnet only.  If
   CLS service.

   The other solutions generally involve a separate VC to carry the subnet is
   excess.  A distinct VC can be set up for each VC supporting a simple point-to-point
   link, GS or
   CLS flow, or, if many flows are aggregated into a single QoS VC, then
   another VC can handle the subnet-specific parts of C and D will account excess traffic for that set of flows.  A VC
   can be set up to handle all excess traffic from the
   link transmission rate and MTU.  An ATM subnet is more complex.

   The edge ingress router will always have an internal packet scheduler, which
   will contribute to C and D.  For this discussion we consider only
   the
   ATM subnet-specific components.  We further assume that egress point.  Since the ATM
   network will be represented as a "pure delay" element, contributing a
   component to D, but not to C.  The reason for this is that C would
   depend on details QoS of the cell scheduling algorithm inside the ATM
   switches, which excess traffic is not known by the edge device, where
   particularly constrained, the AdSpec
   parameters are accumulated.  (In design is quite flexible.  The service
   category for the special case where excess-traffic VC may typically be UBR or ABR,
   although one could use CBR or nrtVBR if the edge
   device does have excess traffic were
   predictable enough information to modify C, it would not know what rate to allocate.  (This wouldn't
   normally be expected for excess traffic, though.)

   Whether a separate VC is used may be
   precluded.)  Generally influenced by the delay behavior design of the whole ATM cloud may
   be expressed abstractly as a fixed constant D_ATM.

   Since
   router scheduler.  The CLS spec suggests two possible
   implementations: one where excess traffic shares the AdSpec values are incremented before any reservation Best Effort
   class scheduler allocation, but at lower priority than other best
   effort traffic.  The other where a separate allocation is
   made, the edge device must have some knowledge about made.  The
   first would allow excess traffic to use the same VC which as normal best
   effort traffic, and the second would be set up in case suggest a reservation were made.  This separate VC.

   TM/UNI 4.0. does not
   really add to the complexity support tagging of the device, since it must also have
   this information traffic in order to make an intelligent VC setup request.
   For example, the edge device may excess of PCR.
   Although UNI 3.x does have a cached table with the
   propagation delay and a reasonable additional delay budget, from
   which it composes a value of CTD separate PCR parameter for the VC setup.  The device may
   learn such information through VC setup negotiation, and, indeed,
   there may be no other way CLP=0 cells
   only, we do not recommend using this feature for reasons of
   interoperability.  This restricts CBR VCs to obtain that information.  However, it
   seems reasonable that these values would use solutions other than
   tagging.  The value of PCR can be cached set higher than necessary for later use when
   new VCs
   conformant traffic, in an effort to support excess traffic on the
   same egress router need to VC.  In some cases this may be established.

   Therefore, we will presume a table with values of MPL (which includes
   propagation delay) and expected queueing delays for each possible
   egress edge device.  (How viable solution, such a table as when
   there is maintained little additional cost imposed for a high PCR.  If PCR can
   be set as high as APB, then the excess traffic is
   implementation specific.) fully accommodated.

3.3 Use of Guaranteed Service Adspec Parameters and Slack Term

   The latter quantity Adspec is simply D_ATM, used by the
   value added Guaranteed Service to the AdSpec D term allow a receiver to account for
   calculate the ATM network.
   When worst-case delay associated with a RESV GS flow.  Three
   quantities, C, D, and MPL, are accumulated (by simple addition of
   components corresponding to each network element) in the PATH message arrives, causing a VC
   from source to receiver.  The resulting delay values can be set up, the requested
   value different
   for CTD should each unique receiver.  The maximum delay is then be given by

           CTD = D_ATM computed as

                  delay <=  b_r/R + MPL C_TOT/R + S_ATM. D_TOT + MPL

   The last term, S_ATM is the portion of the slack term applied to Minimum Path Latency (MPL) includes propagation delay, while
   b_r/R accounts for bursts and C and D include other queueing,
   scheduling and serialization delays.  (We neglect the
   ATM portion effect of
   maximum packet size and peak rate here; see the path.  Recall that the slack term GS specification [8] is positive
   when the receiver can afford
   for a more delay than that computed from the
   AdSpec. detailed equation.)  The ATM edge device may take part (or all) of service rate requested by the
   receiver, R, can be greater than the slack term
   to relax TSpec rate, r_r, resulting in
   lower delay.  The burst size, b_r, is the delay constraint on leaky bucket parameter from
   the ATM VC. receiver TSpec.

   The distribution values of
   delay slack among the nodes C and subnets is network specific.

   An important detail to note is D that a router advertises depend on both the relationship between
   router packet scheduler, and the b/R term characteristics of the (Internet) delay and subnet
   attached to the corresponding MBS/SCR in router.  Each router (or the ATM
   network, when using a VBR VC.  The term b/R accounts source host) takes
   responsibility for its downstream subnet in its advertisement.  For
   example, if the delay
   experienced by the last byte of a burst, of size b, which encounters subnet is a congested node.  In the simple ideal case, where the scheduling
   algorithm emulates a fixed rate server, at rate R, point-to-point link, the delay subnet-
   specific parts of C and D need to account for the
   last byte link transmission
   rate and MTU.  An ATM subnet is b/R.  Once generally more complex.

   For this occurs, discussion, we consider only the stream has been smoothed, ATM subnet-specific
   components, denoted C_ATM and such a delay will not occur at later congested nodes, as long D_ATM.  The ATM network can be
   represented as
   they also serve at a "pure delay" element, where the variable queueing
   delay, given by CVD is captured in D_ATM, and C_ATM = 0.  It is
   possible to use C_ATM only when the ATM service rate equals R.  The form of  This
   may be the delay equation expresses
   this ideal behavior case, for example with C and D acting as error terms.  Now, since a CBR VC with PCR = R.  Usually it
   will be simpler to just advertise the total delay which smooths variation (CDV) in
   D_ATM.

   As discussed in Section 2.6, the burst can occur outside edge router keeps a table with
   values of MPL and D_ATM for each egress router it needs to share VCs
   with.  The values of D_ATM contribute to the ATM cloud, D parameter advertised
   by the b/R term cannot include any delay within edge router, and should accurately reflect the ATM cloud.  However, CDV that the
   router will get in a burst of size MBS VC when it is permitted to enter set up.  Factors that affect CDV,
   such as statistical multiplexing in the ATM network, and it may should be served at a rate no greater than SCR.  We might reasonably expect
   a queueing delay taken
   into account when compiling data for the router's table.  In case of MBS/SCR
   uncertainty, D_ATM can be set to occur at an upper bound.

   When a congested ATM switch.  If RESV message arrives, causing a VC to be set up, the ATM network will impose this delay, then it must requested
   values for CTD and CDV can be included relaxed using the slack term in the value of
   receiver RSpec:

           CTD = D_ATM advertised.  If + MPL + S_ATM
           CDV = D_ATM + S_ATM.

   The term S_ATM is the ATM network can increase its
   bandwidth allocation (e.g., due to CTD being lower than MBS/SCR), portion of the slack term applied to
   decrease this delay, then this behavior should be reflected in the
   value ATM
   portion of D_ATM.  So, the information path.  Recall that the slack term [8] is positive when
   the receiver can afford more delay than that computed from which the
   Adspec.  The ATM edge device
   determines D_ATM must reflect an accurate abstraction may take part (or all) of the actual
   behavior slack term
   S.  The distribution of delay slack among the ATM network.  To the extent that D_ATM nodes and subnets is approximate
   (and it must be an upper bound on the actual delay), it reduces the
   chance that the VC setup will succeed, and/or increases its cost.

4.0 Discussion
   network specific.

   Note that with multipoint VCs the egress edge router may need to
   correct advertised values of C and D.  See discussion in Section 3.1.

4.0 Miscellaneous Items

4.1 Units Conversion

   In

   All rates and token bucket depth parameters that are mapped from IP-
   level parameters to ATM parameters must be corrected for the integrated services domain, effects
   of cell headers, AAL headers and segmentation of packets into cells.
   At the IP layer, bucket sizes depths and rates are measured in bytes and
   bytes/sec, respectively, whereas for ATM, they are measured in cells
   and cells/sec.

   Packets are segmented into 53 byte cells of which the first 5 bytes
   are header information.  For

         B = number of Bytes,
         C = number of cells,

   a rough approximation between the token bucket parameters (rate and
   bucket depth) is

         C = B/48.

   This is actually a lower bound on C and does not take into account
   the extra padding at the end of a partially filled cell, or the 8
   byte trailer in the last cell of an AAL5 encoding.  The actual
   relationship between the number of cells and bytes of one packet is

         C = 1 + int(B/48) + x,

         where x = 1 if B mod 48 > 41
                   0 otherwise.

   where int() is the rounding down operation.  The third term is  0 or
   1 and is 1 only when the remainder of B/48 is 41 or more.   (An
   additional cell is needed because the 41 bytes plus 8 byte trailer
   will not fit in a cell.)

   The above formula is not particularly amenable to engineering
   considerations.  By equating the number of bytes before and after
   segmentation we have

         48 C = B + 8 + A,
   where A is the additional padding used in the last 2 cells and has
   the range 0 <= A <= 47.  From this we obtain a number of  useful
   observations.

   For example, if one believes that the packet lengths are uniformly
   distributed mod 48, then on average, 48 C = B + 8 + 47/2, or C = B/48
   + .65625.

   We can also make use of the upper bound on A to state that 48 C <= B
   + 55.  This is true for any one packet.  Considering the number of
   bytes in a stream of P packets, we have

         48 C <= B + 55 P.

   The number of packets P may not be a readily available quantity.
   However, in terms of the minimum policed unit m, we know that P * m
   <= B.  Hence P <= B/m and 48 C <= B ( 1 + 55/m).  That is,

         C <= B/48 * (1 + 55/m).

5.0 Summary of ATM VC Setup Parameters for Guaranteed Service

   This section describes how to create ATM VCs appropriately matched
   for Guaranteed Service. The key points differentiating among ATM
   choices are that real-time timing is required, that the data flow may
   have a variable rate, and that demotion of non-conforming traffic to
   best effort is required to be in agreement with the definition of GS.
   For this reason, we prefer an rtVBR service in which tagging is
   supported.  Another good match is to use CBR with special handling of
   any non-conforming traffic.

   Note, traffic, usually through another VC, since a CBR
   VC will not accommodate traffic in all cases the excess of PCR.

   Note, these encodings assume point to multipoint connections, where
   the backward channel is not used.  This is done to
   be consistent with rsvp, which generally assumes a multicast
   scenerio.  If a specific situation does not involve multicast, the IP session is unicast only,
   then a point-to-point VC may be used and the IWF may make use of the
   backward channel in a point-to-point VC, channel, provided that the QoS parameters are mapped
   consistently for the service provided.

5.1 Encoding GS Using Real-Time VBR (ATM Forum TM/UNI 4.0)

   AAL
     Type                            5
     Forward CPCS-SDU Size           parameter M of receiver TSpec
     Backward CPCS-SDU Size          0
     SSCS Type                       0 (Null SSCS)
   Traffic Descriptor
     Forward PCR CLP=0+1                                     Note 1
     Backward PCR CLP=0+1            0
     Forward SCR CLP=0                                       Note 1
     Backward SCR CLP=0              0
     Forward MBS (CLP=0)                                     Note 1
     Backward MBS (CLP=0)            0
     BE indicator                    NOT included
     Forward Frame Discard bit       1
     Backward Frame Discard bit      1
     Tagging Forward bit             1 (Tagging requested)
     Tagging Backward bit            1 (Tagging requested)

   Broadband Bearer Capability
     Bearer Class                    16 (BCOB-X)              Note 2
     ATM Transfer Capability         9  (Real time VBR)       Note 3
     Susceptible to Clipping         00 (bit encoding for Not
   susceptible)
     User Plane Configuration        01 (bit encoding for pt-to-mpt)

   Broadband Low Layer Information
     User Information Layer 2
       Protocol                      12 (ISO 8802/2)
     User Information Layer 3
       Protocol                      11 (ISO/IEC TR 9577)    Note 4
      ISO/IEC TR 9577 IPI            204

   QoS Class
     QoS Class Forward               1                       Note 5
     QoS Class Backward              1                       Note 5

   QoS Parameters                                            Note 6
     Acceptable Forward CDV
     Acceptable Forward CLR
     Forward Max CTD

   Note 1:  See discussion Section 2.5.1.
   Note 2:  Value 3 (BCOB-C) can also be used.
   Note 3:  The ATC value 19 is not used.  The value 19 implies CLR
            objective applies to the aggregate  CLP=0+1 stream and
            that does not give desirable treatment of excess
            traffic in the case of IP.
   Note 4:  For QoS VCs supporting GS or CLS, the layer 3 protocol should
            be specified.  For BE VCs, it can be left unspecified, allowing
            the VC to be shared by multiple protocols, following RFC 1755.
   Note 5:  Cf ITU I.365 (Oct 1996) for new definition. QoS Class definitions.
   Note 6:  See section Section 2.6 for the values to be used  The cumulative CDV
            is also provided, but it depends on local implementation, and
            not on values mapped from IP level service parameters. used.

5.2 Encoding GS Using CBR (ATM Forum TM/UNI 4.0)

   It is also possible to support GS using a CBR ``pipe.'' "pipe."   The advantage
   of this is that CBR is probably supported; the disadvantage is that
   data flows may not fill the pipe (utilization loss) and there is no
   tagging option available.

   AAL
     Type                            5
     Forward CPCS-SDU Size           parameter M of receiver TSpec
     Backward CPCS-SDU Size          parameter M of receiver TSpec
     SSCS Type                       0 (Null SSCS)

   Traffic Descriptor
     Forward PCR 0                                           Note 1
     Backward PCR                    0
     Forward PCR 0+1 CLP=0+1                                     Note 1
     Backward PCR 0+1 CLP=0+1            0
     BE indicator                    NOT included
     Forward Frame Discard bit       1
     Backward Frame Discard bit      1
     Tagging Forward bit             1             0 (Tagging not requested)
     Tagging Backward bit            1            0 (Tagging not requested)

   Broadband Bearer Capability
     Bearer Class                    16 (BCOB-X)             Note 2
     ATM Transfer Capability         5  (CBR)                Note 3, 4
     Susceptible to Clipping         00 (bit encoding for Not
   susceptible)
     User Plane Configuration        01 (bit encoding for pt-to-mpt)

   Broadband Low Layer Information
     User Information Layer 2
       Protocol                      12 (ISO 8802/2)
     User Information Layer 3
       Protocol                      11 (ISO/IEC TR 9577)    Note 5
      ISO/IEC TR 9577 IPI            204

   QoS Class
     QoS Class Forward               1                       Note 6
     QoS Class Backward              1                       Note 6

   QoS Parameters                                            Note 7
     Acceptable Forward CDV
     Acceptable Forward CLR
     Forward Max CTD

   Note 1:  See discussion Section 2.5.1.
   Note 2:  Value 1 (BCOB-A) can also be used.
   Note 3:  If bearer class A is chosen the ATC field must be absent.
   Note 4:  The ATC value 7 is not used.  The value 7 implies CLR
            objective applies to the aggregate  CLP=0+1 stream and
            that does not give desirable treatment of excess
            traffic in the case of IP.
   Note 5:  For QoS VCs supporting GS or CLS, the layer 3 protocol should
            be specified.  For BE VCs, it can be left unspecified, allowing
            the VC to be shared by multiple protocols, following RFC 1755.
   Note 6:  Cf ITU I.365 (Oct 1996) for new definition. QoS Class definitions.
   Note 7:  See section Section 2.6 for the values to be used  The cumulative CDV
            is also provided, but it depends on local implementation, and
            not on values mapped from IP level service parameters. used.

5.3 Encoding GS Using Non-Real-Time VBR (ATM Forum TM/UNI 4.0)

   The remaining ATM service categories, including nrtVBR, do not
   provide delay guarantees and cannot be recommended as the best fits.
   However in some circumstances, the best fits may not be available.

   If nrtVBR is used, no hard delay can be given.  However by using a
   variable rate service with low utilization, delay may be
   `reasonable', but not controlled.  The encoding of GS as nrtVBR is
   the same as that for CLS using nrtVBR, except that the Forward PCR
   would be derived from the Tspec TSpec peak rate.  See Section 6.2 below.

5.4 Encoding GS Using ABR (ATM Forum TM/UNI 4.0)

   This

   GS using ABR is a very unlikely combination.  The objective of the
   ABR service is to provide `low' "low" loss rates which, via flow control, can
   result in delays. rates.  The introduction of delays is contrary to the
   design delay objectives of GS. for
   ABR should be expected to be very loose.  If ABR were used for GS,
   the VC parameters would follow as for CLS over ABR.  See Section 6.1.

5.5 Encoding GS Using UBR (ATM Forum TM/UNI 4.0)

   The UBR service is the default lowest common denominator of the
   services.  It cannot provide delay or loss guarantees.  However if it
   is used for GS, it will be encoded in the same way as Best Effort
   over UBR, with the exception that the PCR would be determined from
   the peak rate of the Tspec. receiver TSpec.  See Section 5.1. 7.1.

5.6 Encoding GS Using ATM Forum UNI 3.0/3.1 Specifications

   It is not recommended to support GS using UNI 3.x VBR mode for the
   following reasons.  The Class C bearer class does not represent
   real-time behavior.  Appendix F of UNI 3.1 specification precludes
   the specification of traffic type "VBR" with the timing requirement
   "End to End timing Required" in conjunction with bearer class X.

   It is possible to support GS using a CBR ``pipe.'' "pipe."   The following
   table specifies the support of GS using CBR.

   AAL
     Type                            5
     Forward CPCS-SDU Size           parameter M of receiver TSpec
     Backward CPCS-SDU Size          parameter M of receiver TSpec
     Mode                            1 (Message mode)        Note 1
     SSCS Type                       0 (Null SSCS)

   Traffic Descriptor
     Forward PCR 0 CLP=0                                       Note 2
     Backward PCR CLP=0              0
     Forward PCR 0+1 CLP=0+1                                     Note 2
     Backward PCR 0+1 CLP=0+1            0
     BE indicator                    NOT included
     Tagging Forward bit             1 (Tagging requested)
     Tagging Backward bit            1 (Tagging requested)

   Broadband Bearer Capability
     Bearer Class                    16 (BCOB-X)             Note 3
     Traffic Type                    001 (bit encoding for Constant Bit
   Rate)
     Timing Requirements             01 (bit encoding for Timing
   Required)
     Susceptible to Clipping         00 (bit encoding for Not
   susceptible)
     User Plane Configuration        01 (bit encoding for pt-to-mpt)
   Broadband Low Layer Information
     User Information Layer 2
       Protocol                      12 (ISO 8802/2)
     User Information Layer 3
       Protocol                      11 (ISO/IEC TR 9577)    Note 4
      ISO/IEC TR 9577 IPI            204

   QoS Class
     QoS Class Forward               1
     QoS Class Backward              1

   QoS Parameters
     Parameters are implied by the QOS Class

   Note 1:  Only included for UNI 3.0.
   Note 2:  See discussion, Section 2.5.1. 3.0.
   Note 2:  See discussion, Section 2.5.1.  PCR CLP=0 should be set identical
            to PCR CLP=0+1.  Although this culd potentially allow a CBR VC
            to carry excess traffic as tagged cells, it is not recommended
            since it is not supported in UNI 4.0
   Note 3:  Value 1 (BCOB-A) can also be used. If BCOB-A is used Traffic
            Type and Timing Requirements fields are not included.
   Note 4:  For QoS VCs supporting GS or CLS, the layer 3 protocol should
            be specified.  For BE VCs, it can be left unspecified, allowing
            the VC to be shared by multiple protocols, following RFC 1755.

6.0 Summary of ATM VC Setup Parameters for Controlled Load Service

   This section describes how to create ATM VCs appropriately matched
   for Controlled Load.  CLS traffic is partly delay tolerant and of
   variable rate.  NrtVBR and ABR (for TM/UNI 4.0 only) are the possible best
   choices in supporting CLS.

   Generally we prefer

   Note, these encodings assume point to use point-to-multipoint connections.  However
   this multipoint connections, where
   the backward channel is not yet available in ABR. Other than in ABR, used.  If the encodings
   assume a point-to-multipoint connection.  For a IP session is unicast connection, only,
   then a point-to-point VC may be used and the IWF may make use of the
   backward channel, provided that the QoS parameters would be equal to are mapped
   consistently for the forward parameters. service provided.

6.1 Encoding CLS Using ABR (ATM Forum TM/UNI 4.0)

   AAL
     Type                            5
     Forward CPCS-SDU Size           parameter M of receiver TSpec
     Backward CPCS-SDU Size          parameter M of receiver TSpec
     SSCS Type                       0 (Null SSCS)

   Traffic Descriptor
     Forward PCR CLP=0+1             From line rate                                      Note 1
     Backward PCR CLP=0+1            From line rate            0
     Forward MCR CLP 0+1             From TSpec token bucket rate CLP=0+1                                      Note 1
     Backward MCR CLP 0+1            From TSpec token bucket rate CLP=0+1            0
     BE indicator                    NOT included
     Forward Frame Discard bit       1
     Backward Frame Discard bit      1
     Tagging Forward bit             0 (Tagging not requested)
     Tagging Backward bit            0 (Tagging not requested)

   Broadband Bearer Capability
     Bearer Class                    16 (BCOB-X)              Note 1 2
     ATM Transfer Capability         12 (ABR)
     Traffic Type                    010 (Variable Bit Rate)
     Timing Requirements             10 (Timing Not Required)
     Susceptible to Clipping         00 (Not susceptible)
     User Plane Configuration        00 (For pt-to-pt)

   Broadband Low Layer Information
     User Information Layer 2
       Protocol                      12 (ISO 8802/2)
     User Information Layer 3
       Protocol                      11 (ISO/IEC TR 9577)    Note 2 3
      ISO/IEC TR 9577 IPI            204

   QoS Class
     QoS Class Forward               3               0                       Note 3 4
     QoS Class Backward              3              0                       Note 3 4

   QoS Parameters                                            Note 4 5
     Acceptable Forward CDV
     Acceptable Forward CLR
     Forward Max CTD

   ABR Setup Parameters                                      Note 5 6
   ABR Additional Parameters                                 Note 5 6

   Note 1:  See discussion, Section 2.5.2.
   Note 2:  Value 3 (BCOB-C) can also be used.

   Note 2: 3:  For QoS VCs supporting GS or CLS, the layer 3 protocol should
            be specified.  For BE VCs, it can be left unspecified, allowing
            the VC to be shared by multiple protocols, following RFC 1755.
   Note 3: 4:  Cf ITU I.365 (Oct 1996) for new definition. QoS Class definitions.
   Note 4: 5:  See section Section 2.6 for the values to be used.  The cumulative CDV
   Note 6:  Discussion of ABR-specific parameters is also provided, but it depends beyond the scope of
            this document.  These generally depend on local implementation, implementation and
            not on values mapped from IP level service parameters.
   Note 5:  Discussion of these parameters is beyond (with the scope
            exception of this draft. MCR).

6.2 Encoding CLS Using Non-Real-Time VBR (ATM Forum TM/UNI 4.0)

   AAL
     Type                            5
     Forward CPCS-SDU Size           parameter M of receiver TSpec
     Backward CPCS-SDU Size          0          parameter M of receiver TSpec
     SSCS Type                       0 (Null SSCS)

   Traffic Descriptor
     Forward PCR CLP=0+1             From line rate                                     Note 1
     Backward PCR CLP=0+1            0
     Forward SCR CLP=0               From TSpec token bucket rate                                       Note 1
     Backward SCR CLP=0              0
     Forward MBS (CLP=0)             From TSpec bucket size param                                     Note 1
     Backward MBS (CLP=0)            0
     BE indicator                    NOT included
     Forward Frame Discard bit       1
     Backward Frame Discard bit      1
     Tagging Forward bit             1 (Tagging requested)
     Tagging Backward bit            1 (Tagging requested)

   Broadband Bearer Capability
     Bearer Class                    16 (BCOB-X)             Note 1 2
     ATM Transfer Capability         10 (Non-real time VBR)  Note 2, 3 3, 4
     Susceptible to Clipping         00 (bit encoding Not susceptible)
     User Plane Configuration        01 (bit encoding pt-to-mpt)

   Broadband Low Layer Information
     User Information Layer 2
       Protocol                      12 (ISO 8802/2)
     User Information Layer 3
       Protocol                      11 (ISO/IEC TR 9577)    Note 4 5
      ISO/IEC TR 9577 IPI            204
   QoS Class
     QoS Class Forward               3                       Note 5 6
     QoS Class Backward              3                       Note 5 6

   QoS Parameters                                            Note 6 7
     Acceptable Forward CDV
     Acceptable Forward CLR
     Forward Max CTD

   Note 1:  See discussion, Section 2.5.2.
   Note 2:  Value 3 (BCOB-C) can also be used.
   Note 2: 3:  If bearer class C is used, the ATC field must be absent
   Note 3: 4:  The ATC value 11 is not used.  The value 11 implies CLR
            objective applies to the aggregate  CLP=0+1 stream and
            that does not give desirable treatment of excess
            traffic in the case of IP.
   Note 4: 5:  For QoS VCs supporting GS or CLS, the layer 3 protocol should
            be specified.  For BE VCs, it can be left unspecified, allowing
            the VC to be shared by multiple protocols, following RFC 1755.
   Note 5: 6:  Cf ITU I.365 (Oct 1996) for new definition.
   Note 6:  See section 2.6 for the values to be used.  The cumulative CDV
            is also provided, but it depends on local implementation, and
            not on QoS Class definitions.
   Note 7:  See Section 2.6 for the values mapped from IP level service parameters. to be used.

6.3 Encoding CLS Using Real-Time VBR (ATM Forum TM/UNI 4.0)

   The encoding of CLS using rtVBR imposes a hard limit on the delay,
   which is specified as an end-to-end delay in the ATM network.  This
   is more stringent than the CLS service specifies and may result in
   less utilization of the network. requires.

   If rtVBR is used to encode CLS, then the encoding is essentially the
   same as that for GS.  The exceptions are that the Forward PCR is
   derived from the line rate and probably a different value of the
   transit delay and CDV will be specified.  See Section 3.1. 5.1 and discussion in Section
   2.5.2.

6.4 Encoding CLS Using CBR (ATM Forum TM/UNI 4.0)

   The encoding of CLS using

   Although CBR is more stringent than using rtVBR
   since it does not explicitly take into account the variable rate
   of the data.
   Consequently there source data, it may be even lower utilization of the network. convenient to use ATM connectivity between
   edge routers to provide a simple "pipe" service, as a leased line
   replacement.

   To use CBR for CLS, the same encoding as in Section 3.2 for GS over CBR (Section 5.2
   would be used.  However a different set of values of the QoS parameters will
   likely be used.  See Section 2.5.2.

6.5 Encoding CLS Using UBR (ATM Forum TM/UNI 4.0)

   This encoding gives no QoS guarantees and would be done guarantees.  If used, it is coded in the
   same way as for BE traffic. over UBR, except that the PCR would be determined
   from the peak rate of the receiver TSpec.  See Section 5.1. 7.1.

6.6 Encoding CLS Using Non-Real-Time VBR as in ATM Forum UNI 3.0/3.1 Specifications

   This encoding is equivalent to the nrtVBR service category.

   AAL
     Type                            5
     Forward CPCS-SDU Size           parameter M of receiver TSpec
     Backward CPCS-SDU Size          0
     Mode                            1 (Message mode)        Note 1
     SSCS Type                       0 (Null SSCS)

   Traffic Descriptor
     Forward PCR CLP=0+1             From line rate                                     Note 2
     Backward PCR CLP=0+1            0
     Forward SCR CLP=0               From TSpec token bucket rate                                       Note 2
     Backward SCR CLP=0              0
     Forward MBS (CLP=0)             From TSpec bucket size param                                     Note 2
     Backward MBS (CLP=0)            0
     BE indicator                    NOT included
     Tagging Forward bit             1 (Tagging requested)
     Tagging Backward bit            1 (Tagging requested)

   Broadband Bearer Capability
     Bearer Class                    16 (BCOB-X)             Note 2 3
     Traffic Type                    010 (bit encoding for Variable Bit
   Rate)
     Timing Requirements             00 (bit encoding for No Indication)
     Susceptible to Clipping         00 (bit encoding for Not
   susceptible)
     User Plane Configuration        01 (bit encoding for For pt-to-mpt)

   Broadband Low Layer Information
     User Information Layer 2
       Protocol                      12 (ISO 8802/2)
     User Information Layer 3
       Protocol                      11 (ISO/IEC TR 9577)    Note 3 4
      ISO/IEC TR 9577 IPI            204

   QoS Class
     QoS Class Forward               3
     QoS Class Backward              3

   QoS Parameters
     Parameters are implied by the QOS Class

   Note 1:  Only included for UNI 3.0.
   Note 2:  See discussion, Section 2.5.2.
   Note 3:  Value 3 (BCOB-C) can also be used. If BCOB-C is used Traffic
            Type and Timing Requirements fields are not included.
   Note 3: 4:  For QoS VCs supporting GS or CLS, the layer 3 protocol should
            be specified.  For BE VCs, it can be left unspecified, allowing
            the VC to be shared by multiple protocols, following RFC 1755.

7.0 Summary of ATM VC Setup Parameters for Best Effort Service

This section describes how to create should be considered informational only.  RFC 1755 [10] and
the IETF ION working group draft on ATM VCs appropriately matched signalling support for IP over
ATM using UNI 4.0 [11] provide more definitive specification of Best
Effort IP service over ATM.

The best-matched ATM service category to IP Best Effort. Effort is UBR.  We
provide the setup details for this case below.  The BE service does not need a not
require reservation of resources.

The following subsections

Note, VCs supporting best effort service are for information only.  See usually point to point,
rather than point to multipoint, and the IETF ION
working group draft backward channels of VCs are
used.  In specific cases where VCs are set up to support best effort
multicast sessions, multipoint VCs can be used and the backward channels
would be not have resources reserved.  Related situations include
transport of excess traffic on ATM signalling IP-multicast QoS sessions, or to support for IP over ATM using UNI
4.0 [11] for recommendations.
the subset of multicast end systems that have not made rsvp
reservations.

7.1 Encoding Best Effort Service Using UBR (ATM Forum TM/UNI 4.0)

   This section is for information only.  For recommendation, see the
   IETF ION working group draft on ATM signalling support for IP over
   ATM using UNI 4.0 [11].
   AAL
     Type                            5
     Forward CPCS-SDU Size           9180 (default MTU of link for AAL5)
     Backward CPCS-SDU Size          9180 (default MTU of link for AAL5)
     SSCS Type                       0 (Null SSCS)

   Traffic Descriptor
     Forward PCR CLP=0+1             From line rate                                         Note 1
     Backward PCR CLP=0+1            0
     BE indicator                    included
     Forward Frame Discard bit       1
     Backward Frame Discard bit      1
     Tagging Forward bit             1 (Tagging requested)
     Tagging Backward bit            1 (Tagging requested)

   Broadband Bearer Capability
     Bearer Class                    16 (BCOB-X)                 Note 1 2
     ATM Transfer Capability         10 (Non-real time VBR)      Note 2 3
     Susceptible to Clipping         00 (bit encoding for Not susceptible)
     User Plane Configuration        01 (bit encoding for pt-to-mpt)

   Broadband Low Layer Information
     User Information Layer 2
       Protocol                      12 (ISO 8802/2)
     User Information Layer 3
       Protocol                      11 (ISO/IEC TR 9577)    Note 3 4
      ISO/IEC TR 9577 IPI            204

   QoS Class
     QoS Class Forward               0
     QoS Class Backward              0

   Note 1:  See discussion, Section 2.5.3.
   Note 2:  Value 3 (BCOB-C) can also be used.
   Note 2: 3:  If bearer class C is used, the ATC field must be absent
   Note 3: 4:  For QoS VCs supporting GS or CLS, the layer 3 protocol should
            be specified.  For BE VCs, it can be left unspecified, allowing
            the VC to be shared by multiple protocols, following RFC 1755.
    .fi

7.2 Encoding Best Effort Service Using Other ATM Service Categories

See the IETF ION working group draft on ATM signalling support for IP
over ATM using UNI 4.0 [11].

8.0 Security

   Some security issues are raised in the rsvp specification [2], which
   would apply here as well.  There are no additional security
   considerations raised in this document.

9.0 Acknowledgements

   The authors would like to thank the members of the ISSLL working
   group for their input. In particular, thanks to Drew Perkins and Jon
   Bennett of Fore Systems, Roch Guerin of IBM and IBM, Susan Thomson and Sudha
   Ramesh of Bellcore.

Appendix 1  Abbreviations

       AAL           ATM Adaptation Layer
       ABR           Available Bit Rate
       APB           Available Path Bandwidth (int-serv GCP)
       ATM           Asynchronous Transfer Mode
       B-LLI         Broadband Low Layer Information
       BCOB          Broadband Connection-Oriented Bearer Capability
       BCOB-{A,C,X}  Bearer Class A, C, or X
       BE            Best Effort
       BT            Burst Tolerance
       CBR           Constant Bit Rate
       CDV           Cell Delay Variation
       CDVT          Cell Delay Variation Tolerance
       CLP           Cell Loss Priority (bit)
       CLR           Cell Loss Ratio
       CLS           Controlled Load Service
       CPCS          Common Part Convergence Sublayer
       CTD           Cell Transfer Delay
       EOM           End of Message
       FFS           For Further Study
       GCP           General Characterization Parameter
       GCRA          Generic Cell Rate Algorithm
       GS            Guaranteed Service
       IE            Information Element
       IETF          Internet Engineering Task Force
       IP            Internet Protocol
       IPI           Initial Protocol Identifier
       IS            Integrated Services
       ISSLL         Integrated Services over Specific Link Layers
       ITU           International Telecommunication Union
       IWF           Interworking Function
       LIJ           Leaf Initiated Join
       LLC           Logical Link Control
       MBS           Maximum Burst Size
       MCR           Minimum Cell Rate
       MPL           Minimum Path Latency
       MTU           Maximum Transfer Unit
       nrtVBR        Non-real-time VBR
       PCR           Peak Cell Rate
       PDU           Protocol Data Unit
       PVC           Permanent Virtual Connection
       QoS           Quality of Service
       RESV          Reservation Message (of rsvp protocol)
       RFC           Request for Comment
       RSVP          Resource Reservation Protocol
       Rspec         Reservation Specification
       rtVBR         Real-time VBR
       SCR           Sustained Cell Rate
       SDU           Service Data Unit
       SIG           ATM Signaling (ATM Forum document)
       SNAP          Subnetwork Attachment Point
       SSCS          Service-Specific Convergence Sub-layer
       SVC           Switched Virtual Connection
       Sw            Switch
       TCP           Transport Control Protocol
       TM            Traffic Management
       TSpec         Traffic Specification
       UBR           Unspecified Bit Rate
       UNI           User-Network Interface
       UPC           Usage Parameter Control (ATM traffic policing function)
       VBR           Variable Bit Rate
       VC            (ATM) Virtual Connection

REFERENCES

References

   [1]  R. Braden, D. Clark and S. Shenker, "Integrated Services in the
        Internet Architecture: an Overview", RFC 1633, June 1994.

   [2]  R. Braden, L. Zhang, S. Berson, S. Herzog and S. Jamin,
        "Resource ReSerVation Protocol (RSVP) - Version 1 Functional
        Specification", Internet Draft, May 1996, <draft-ietf-rsvp-
        spec-12.txt>

   [3]  The ATM Forum, "ATM User-Network Interface Specification, Ver-
        sion 3.0", Prentice Hall, Englewood Cliffs NJ, 1993.

   [4]  The ATM Forum, "ATM User-Network Interface Specification, Ver-
        sion 3.1", Prentice Hall, Upper Saddle River NJ, 1995.

   [5]  The ATM Forum, "ATM User-Network Interface (UNI) Signalling
        Specification, Version 4.0", Prentice Hall, Upper Saddle River
        NJ, specification finalized July 1996; expected publication,
        late 1996; available at ftp://ftp.atmforum.com/pub. ftp://ftp.atmforum.com/pub/approved-
        specs/af-sig-0061.000.ps.

   [6]  The ATM Forum, "ATM Traffic Management Specification, Version
        4.0", Prentice Hall, Upper Saddle River NJ; specification final-
        ized April 1996; expected publication, late 1996; available at
        ftp://ftp.atmforum.com/pub.
        ftp://ftp.atmforum.com/pub/approved-specs/af-tm-0056.000.ps.

   [7]  M. W. Garrett, "A Service Architecture for ATM: From Applica-
        tions to Scheduling", IEEE Network Mag., Vol. 10, No. 3, pp. 6-
        14, May 1996.

   [8]  S. Shenker, C. Partridge and R. Guerin, "Specification of
        Guaranteed Quality of Service", Internet Draft, August 1996,
        <draft-ietf-intserv-guaranteed-svc-06.txt> July 1997,
        <draft-ietf-intserv-guaranteed-svc-08.txt>

   [9]  J. Wroclawski, "Specification of the Controlled-Load Network
        Element Service", Internet Draft, August November 1996, draft-ietf-
        intserv-ctrl-load-svc-03.txt <draft-ietf-
        intserv-ctrl-load-svc-04.txt>

   [10] M. Perez, F. Liaw, A. Mankin, E. Hoffman, D. Grossman and A.
        Malis, "ATM Signaling Support for IP over ATM", RFC 1755, Febru-
        ary 1995.

   [11] M. Perez and A. Mankin, "ATM Signalling Signaling Support for IP over ATM -
        UNI 4.0 Update", Internet Draft, November 1996, <draft-ietf-
        ion-sig-uni4.0-01.txt> May 1997, <draft-ietf-ion-sig-
        uni4.0-04.txt>

   [12] S. Berson, L. Berger, "IP Integrated Services with RSVP over
        ATM", Internet Draft, September 1996, <draft-ietf-issll-atm-
        support-01.txt>

   [13] S. Shenker and J. Wroclawski, "Network Element Service Specifi-
        cation Template", Internet Draft, November 1995, <draft-ietf-
        intserv-svc-template-02.txt>

   [14] J. Wroclawski, "The Use of RSVP with IETF Integrated Services",
        Internet Draft, August 1996, <draft-ietf-intserv-use-00.txt>

   [15] M. Borden, E. Crawley, B. Davie and S. Batsell, "Integration of
        Real-time Services in an IP-ATM Network Architecture", "IP
        Authentication Header", RFC 1821, August 1995.

   [16] J. Heinanen, "Multiprotocol Encapsulation over ATM Adaptation
        Layer 5", RFC 1483, July 1993.

AUTHORS' ADDRESSES

   [17] M. Laubach, "Classical IP and ARP over ATM", RFC 1577, January
        1994.

   [18] L. Berger, "RSVP over ATM Implementation Requirements", Internet
        Draft, July 1997, <draft-ietf-issll-atm-imp-req-00.txt>

   [19] L. Berger, "RSVP over ATM Implementation Guidelines", Internet
        Draft, July 1997, <draft-ietf-issll-imp-guide-01.txt>

   [20] A. Romanow, S. Floyd, "Dynamics of TCP Traffic over ATM Net-
        works", IEEE J. Sel. Areas in Commun., Vol. 13, No. 4, pp. 633-
        -41, May 1995,

   [21] S. Floyd, V. Jacobson, "Link-sharing and Resource Management
        Models for Packet Networks", IEEE/ACM Trans. Networking, Vol. 3,
        No. 4, August 1995.

   [22] S. Shenker and J. Wroclawski, "General Characterization Parame-
        ters for Integrated Service Network Elements", Internet Draft,
        July 1997, <draft-ietf-intserv-charac-03.txt>

Authors' Addresses

Mark W. Garrett                  Marty Borden
Bellcore                         New Oak Communications, Inc.
445 South Street                 42 Nanog Nagog Park
Morristown, NJ 07960             Acton MA, 01720
USA                              USA

phone: +1 201 829-4439           phone: +1 508 266-1011
email: mwg@bellcore.com          email: mborden@newoak.com

Table of Contents