Skip to main content

Path Computation Based on Precision Availability Metrics
draft-contreras-pce-pam-07

Document Type Active Internet-Draft (individual)
Authors Luis M. Contreras , Fernando Agraz , Salvatore Spadaro , Quan Xiong
Last updated 2026-07-06
RFC stream (None)
Intended RFC status (None)
Formats
Stream Stream state (No stream defined)
Consensus boilerplate Unknown
RFC Editor Note (None)
IESG IESG state I-D Exists
Telechat date (None)
Responsible AD (None)
Send notices to (None)
draft-contreras-pce-pam-07
PCE                                                      L. M. Contreras
Internet-Draft                                                Telefonica
Intended status: Informational                                  F. Agraz
Expires: 7 January 2027                                       S. Spadaro
                                    Universitat Politecnica de Catalunya
                                                                Q. Xiong
                                                         ZTE Corporation
                                                             6 July 2026

        Path Computation Based on Precision Availability Metrics
                       draft-contreras-pce-pam-07

Abstract

   This document extends PCEP to support Precision Availability Metrics
   (PAM) [RFC9544] for path computation.  The optimization objectives
   for PAM computations are encoded using the Objective Function (OF)
   object defined in [RFC5541], allowing PCCs to specify precise
   optimization criteria for services with SLO requirements.  And a PCE
   can report the statistical characterization associated with a
   computed path.

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at https://datatracker.ietf.org/drafts/current/.

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

   This Internet-Draft will expire on 7 January 2027.

Copyright Notice

   Copyright (c) 2026 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents (https://trustee.ietf.org/
   license-info) in effect on the date of publication of this document.

Contreras, et al.        Expires 7 January 2027                 [Page 1]
Internet-Draft                PAM-based PCE                    July 2026

   Please review these documents carefully, as they describe your rights
   and restrictions with respect to this document.  Code Components
   extracted from this document must include Revised BSD License text as
   described in Section 4.e of the Trust Legal Provisions and are
   provided without warranty as described in the Revised BSD License.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   2
   2.  Terminology . . . . . . . . . . . . . . . . . . . . . . . . .   3
   3.  Rationale of the usage of PAM for path calculation  . . . . .   3
     3.1.  Dynamic behavior of performance parameters  . . . . . . .   3
     3.2.  Applicability . . . . . . . . . . . . . . . . . . . . . .   4
     3.3.  Usage of collected metrics  . . . . . . . . . . . . . . .   4
     3.4.  Calculation or selection of the path  . . . . . . . . . .   6
   4.  PAM Objective Functions . . . . . . . . . . . . . . . . . . .   7
     4.1.  PAM Compliance Objective Function . . . . . . . . . . . .   8
     4.2.  Minimum Violated Intervals Objective Function . . . . . .   9
     4.3.  Minimum Severe Violated Intervals Objective Function  . .  10
     4.4.  Interaction with Path Constraints . . . . . . . . . . . .  11
   5.  PAM report TLV  . . . . . . . . . . . . . . . . . . . . . . .  11
   6.  Security and operational considerations . . . . . . . . . . .  12
     6.1.  Security considerations . . . . . . . . . . . . . . . . .  12
     6.2.  Operational considerations  . . . . . . . . . . . . . . .  12
   7.  IANA Considerations . . . . . . . . . . . . . . . . . . . . .  12
     7.1.  New PCE Objective Function Codes  . . . . . . . . . . . .  12
     7.2.  PCEP TLV  . . . . . . . . . . . . . . . . . . . . . . . .  13
   8.  Informative References  . . . . . . . . . . . . . . . . . . .  13
   Appendix A.  Path Histogram Composition . . . . . . . . . . . . .  15
     A.1.  Additive Metrics  . . . . . . . . . . . . . . . . . . . .  15
     A.2.  Multiplicative Metrics  . . . . . . . . . . . . . . . . .  15
     A.3.  Maximization / Minimization Metrics (Bottleneck
           Metrics)  . . . . . . . . . . . . . . . . . . . . . . . .  15
   Acknowledgements  . . . . . . . . . . . . . . . . . . . . . . . .  16
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  16

1.  Introduction

   Some network services, such as Network Slices [RFC9543] and
   Deterministic Networking [RFC8578] [RFC8655], express their
   performance requirements using Service Level Objectives (SLOs).

   At the time of calculating a path by the PCE, the METRIC object
   [RFC5440] serves for the purposes of indicating either the metric
   that MUST be optimized by the path computation algorithm, or a bound
   on the path cost that MUST NOT be exceeded for the path to be
   considered as acceptable.  The value of the metric refers to the
   instantaneous observed behavior of that parameter, without a notion

Contreras, et al.        Expires 7 January 2027                 [Page 2]
Internet-Draft                PAM-based PCE                    July 2026

   of behavior along the preceding time.  This cannot be sufficient for
   certain networking services which require to experience stable
   behavior along the time according to their SLOs.

   Precision Availability Metrics (PAM) [RFC9544] introduce statistical
   performance attributes, including Violated Intervals (VI), Severe
   Violated Intervals (SVI), Violated Interval Ratio (VIR), and Severely
   Violated Interval Ratio (SVIR).  These metrics describe the
   historical or probabilistic behavior of a path across observation
   intervals, rather than its instantaneous state.

   This document extends PCEP to support PAM-based path computation.  It
   defines:

   *  PAM-specific Objective Functions (OFs), encoded using the OF
      object defined in [RFC5541], to instruct the PCE on how to rank
      and select paths based on statistical SLO compliance.

   *  PAM Report TLVs, used by the PCE to convey the statistical
      characterization of the computed path to the Path Computation
      Client (PCC).

2.  Terminology

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

   In addition, the terms defined in [RFC9544] are also used in this
   document.

3.  Rationale of the usage of PAM for path calculation

3.1.  Dynamic behavior of performance parameters

   [RFC9544] introduced the concept of intervals for measuring the
   behavior of measurable performance parameters against some predefined
   thresholds.  Those intervals consider a given time window.  Thus, it
   is possible to define a Violated Interval (VI) as the time interval
   during which at least one of the performance parameters presents
   degradation respect to a predefined optimal level threshold.
   Similarly, when the threshold is defined as critical, the degradation
   of the performance parameter in a time window generates a Severe
   Violated Interval (SVI).

   Taking into account the VIs and SVIs it is feasible to generate
   availability metrics showing some degree of historic behavior in the
   form of the following ratios:

Contreras, et al.        Expires 7 January 2027                 [Page 3]
Internet-Draft                PAM-based PCE                    July 2026

   *  Violated Interval Ratio (VIR), defined as the ratio of the summed
      numbers of VIs and SVIs to the total number of time unit intervals
      along a predefined availability period.

   *  Severely Violated Interval Ratio (SVIR), defined as the ratio of
      SVIs to the total number of time unit intervals along a predefined
      availability period.

   At the time of provisioning a networking service which requires
   stable SLOs along the time, it is important to ensure that the
   selected path has shown such stable behavior in the past.  Despite
   the fact that the past behavior is not a guarantee of future
   behavior, it can be presumed that those paths with lower VIR and SVIR
   will better satisfy the SLOs of the intended networking service.
   Alternatively, PAM can be used by the path computation entity for
   fine-grained path computation.  Then PAM are useful criteria for
   calculating and selecting paths.

3.2.  Applicability

   Three situations of applicability of precision metrics can be
   identified:

   *  The provision of a path according to the desired behavior along
      the time.  In this scenario different segments of a potential path
      could be monitored before the path is created.  The path
      calculation can take into consideration the measured
      characteristics of the segments forming that path for decision.

   *  The selection of a path according to its long-run characteristics.
      In this scenario, an existing path being monitored along the time
      can be selected if its behavior is compliant with the long-run
      behavior expected by the customer.

   *  The triggering of corrective actions for a selected path.  It
      could be the case that a selected path suffers degradation.  The
      precision metrics can assist on the identification of such
      potential problems, e.g, raising incidents or anomalies to
      operational groups, as described in
      [I-D.ietf-nmop-network-incident-yang].

3.3.  Usage of collected metrics

   The Traffic Engineering Database (TED) defined in [RFC4655] could be
   considered as the component providing the precision metrics of
   interest.

Contreras, et al.        Expires 7 January 2027                 [Page 4]
Internet-Draft                PAM-based PCE                    July 2026

   The TED stores information related to the network topology, including
   nodes, links, link attributes (e.g., bandwidth, delay), and any
   constraints relevant for traffic engineering.  It is dynamically
   updated with information received via routing protocols (e.g., OSPF-
   TE, IS-IS-TE), ensuring the PCE has up-to-date knowledge of the
   network status and behavior.  It is also possible to define policies
   like administrative group (coloring), to be used in constraint-based
   path computation.

   In order to support precision metrics, the TED could be extended to
   support e.g. time-series storage and processing capabilities (e.g.,
   to derive histograms from them, as described for instance in
   Appendix A).  The metrics could be gathered from in-band telemetry,
   active probing mechanisms, or streaming telemetry via standardized
   interfaces, as complementary information sources to the information
   received from routing protocols.

   Assuming that capability, the PCE queries the TED for compliance with
   precision constraints.

                                                  - Topology
             Path computation request             - Single value metrics
                based on PAM metrics              - Precision metrics
    +-------------+         +-------------+         +-------------+
    |     Path    |         |     Path    |         |   Traffic   |
    | Computation |<------->| Computation |<------->| Engineering |
    |    Client   |         |   Element   |         |  Database   |
    +-------------+         +-------------+         +-------------+
                                                           ^
                                                           |
                                                           v
                                                    +-------------+
                                                    |    Data     |
                                                    |   Sources   |
                                                    +-------------+
                                                 - Link state info
                                                 - Active Probes
                                                 - Streaming telemetry
                                                 - In-band OAM
                                                 - etc

             Figure 1: Usage of precision metrics stored in TED

   The implementation of the TED and its support to the collection,
   processing and generation of the precision metrics is out of scope of
   this document.  Similarly, the mechanisms used to collect telemetry
   and build the statistical characterisation are outside the scope of
   this document.

Contreras, et al.        Expires 7 January 2027                 [Page 5]
Internet-Draft                PAM-based PCE                    July 2026

3.4.  Calculation or selection of the path

   For a given metric, i.e. metric X, it is defined a frequency of
   values per bin for such a metric (e.g., if the metric refers to
   latency, a way of expressing it could be to consider the latency
   below 20 ms the 90% of the time, and below 25 ms the 99% of the
   time).  Thus, the calculation or selection of a path for such a
   metric X will consist on the comparison of the frequency of the
   metric values per bin, so that the intended path behaves equal or
   better than such described behavior.  For that purpose, the
   statistical behavior of the path is characterized e.g. as described
   in Appendix A.

   When selecting a path, the PCE evaluates candidate paths according to
   the following procedure:

   1.  Retrieve PAM information associated with links and path segments.

   2.  Build the end-to-end statistical characterization of each
       candidate path.

   3.  Apply mandatory PAM constraints to eliminate infeasible paths and
       rank feasible ones.

   4.  Apply PAM objective functions across the remaining feasible
       candidate paths.

   5.  Select the path maximizing the objective function value and
       report back the statistical characterization associated with the
       selected path by means of a PAM Report TLV..

Contreras, et al.        Expires 7 January 2027                 [Page 6]
Internet-Draft                PAM-based PCE                    July 2026

                             +---------------------+
                             | Candidate Paths     |
                             +----------+----------+
                                        |
                                        v
                             +---------------------+
                             | PAM Constraints     |
                             | Evaluation          |
                             +----------+----------+
                                        |
                                 Feasible Paths
                                        |
                                        v
                             +---------------------+
                             | PAM Objective       |
                             | Function            |
                             +----------+----------+
                                        |
                                        v
                             +---------------------+
                             | Selected Path       |
                             +---------------------+

                             Figure 2: Process

   The construction of end-to-end statistical distributions derived from
   the per-link observations is implementation dependent and outside the
   scope of this document.  The resulting end-to-end distribution SHALL
   be represented using the PAM TLVs defined by this document.

4.  PAM Objective Functions

   The Path Computation Element (PCE) can use Precision Availability
   Metrics (PAM) in two different ways during path computation.  First,
   PAM information can be used as a path constraint.  In this mode,
   candidate paths that do not satisfy the requested PAM characteristics
   are discarded during the path computation process.

   Second, PAM information can be used as an optimization criterion.  In
   this case, multiple candidate paths can satisfy the requested PAM
   constraints and the PCE uses one or more PAM-related objective
   functions to identify the preferred path among the feasible
   alternatives.  The specific algorithms used by a PCE implementation
   are outside the scope of this document.

   This section defines PAM objective functions that extend the OF
   object defined in [RFC5541].  These objective functions enable PAM-
   aware path computation and selection.

Contreras, et al.        Expires 7 January 2027                 [Page 7]
Internet-Draft                PAM-based PCE                    July 2026

4.1.  PAM Compliance Objective Function

   The PAM Compliance Objective Function aims at selecting the candidate
   path whose statistical behavior exhibits the highest level of
   compliance with the requested Service Level Objective (SLO).

   When multiple feasible paths are available, the PCE evaluates the
   statistical characterization associated with each path and selects
   the one providing the highest degree of compliance with the requested
   PAM profile.

   The method used to determine the level of compliance is
   implementation-specific.

   This objective function is particularly useful when the PCC requests
   a specific statistical distribution describing the expected service
   behavior over a given observation interval.

   The description of the new objective function is as follows.

   *  Objective Function Code: TBD1

   *  Name: PAM Compliance (PAM-COMP)

   *  Description: Selects the candidate path with highest compliance to
      requested PAM characteristics.  The PCE evaluates statistical
      distributions and selects the path providing optimal alignment
      with requested SLO profiles.

   The objective function is formulated using the following terminology:

   *  A network comprises a set of N links {Li, (i=1...N)}.

   *  A path P is a list of K links {Lpi,(i=1...K)}.

   *  Metric of link L is denoted M(L).  This can be any metric such as
      path delay, path delay variation, or path loss as per [RFC8233].

   *  For a given metric X, the requested PAM profile is denoted R_X.
      R_X is composed of a set of bins {Bj, (j=1...B)} and the
      corresponding requested frequency values {rj, (j=1...B)}. Each
      pair (Bj, rj) describes the expected statistical behavior of the
      metric over the observation interval.  For example, for a delay
      metric, a bin may indicate that the path delay is expected to be
      below a given threshold for at least a specified fraction of the
      observation interval.

Contreras, et al.        Expires 7 January 2027                 [Page 8]
Internet-Draft                PAM-based PCE                    July 2026

   *  The statistical characterization of metric X for a candidate path
      P is denoted H_X(P).  H_X(P) is composed of the same set of bins
      {Bj, (j=1...B)} and the corresponding observed, estimated, or
      derived frequency values {pj(P), (j=1...B)} for the candidate
      path.

   *  For each bin Bj, the compliance of path P with respect to the
      requested PAM profile R_X is denoted c_j(P,R_X).  The definition
      of c_j(P,R_X) depends on the semantics of the metric and on
      whether lower or higher values are preferable.  For metrics where
      lower values are preferable, such as delay, a path is considered
      more compliant when the frequency of values not exceeding the
      requested bin threshold is greater than or equal to the requested
      frequency.  For metrics where higher values are preferable, the
      comparison is applied in the opposite direction.

   *  The overall PAM compliance of a path P with respect to the
      requested PAM profile R_X is denoted C(P,R_X), where C(P,R_X) =
      min { c_j(P,R_X), j=1...B }. This formulation captures the weakest
      compliance level across all the requested bins and therefore
      favors paths whose statistical behavior is aligned with the
      complete requested PAM profile, rather than with only a subset of
      the requested bins.

   *  The PAM-COMP OF is to find a path P such that C(P,R_X) is
      maximized among the feasible candidate paths through Maximize
      C(P,R_X), subject to the path constraints applicable to the
      computation.

   *  A path P is fully compliant with the requested PAM profile R_X
      when all the requested bin-level conditions are satisfied.  When
      more than one fully compliant path exists, the PCE SHOULD prefer
      the path with the highest value of C(P,R_X).  When no candidate
      path can fully satisfy the requested PAM profile and the PAM
      profile is used as a desirable optimization criterion rather than
      as a mandatory constraint, the PCE MAY select the path with the
      highest value of C(P,R_X) and SHOULD report the resulting
      statistical characterization using the PAM Report TLV.

4.2.  Minimum Violated Intervals Objective Function

   The Minimum Violated Intervals (Min-VI) Objective Function aims at
   selecting the path exhibiting the lowest occurrence of Violated
   Intervals (VI).  When several candidate paths satisfy the requested
   constraints, the PCE SHOULD prefer the path associated with the
   lowest VI occurrence.  This objective function favors paths
   presenting fewer periods of performance degradation with respect to
   the requested service objectives.

Contreras, et al.        Expires 7 January 2027                 [Page 9]
Internet-Draft                PAM-based PCE                    July 2026

   The description of the new objective function is as follows:

   *  Objective Function Code: TBD2

   *  Name: Minimum Violated Intervals (MIN-VI)

   *  Description: Selects the path with lowest Violated Interval (VI)
      occurrence, minimizing periods of performance degradation relative
      to SLO thresholds.

   The objective function is formulated using the following terminology:

   *  A network comprises a set of N links {Li, (i=1...N)}.

   *  A path P is a list of K links {Lpi,(i=1...K)}.

   *  Violated Interval on link L is denoted VI(L)

   *  The Violated Intervals of a path P is denoted VI(P), where VI(P) =
      Max {VI(Lpi), (i=1...K)}.

   *  The Min-VI OF is to find a path P such that ( Max { VI(Lpi),
      i=1...K } ) is minimized.

4.3.  Minimum Severe Violated Intervals Objective Function

   The Minimum Severe Violated Intervals (Min-SVI) Objective Function
   aims at selecting the path exhibiting the lowest occurrence of Severe
   Violated Intervals (SVI).  When several candidate paths satisfy the
   requested constraints, the PCE SHOULD prefer the path associated with
   the lowest SVI occurrence.  This objective function is particularly
   relevant for services where severe service degradation events must be
   minimized.

   The description of the new objective function is as follows:

   *  Objective Function Code: TBD3

   *  Name: Minimum Severe Violated Intervals (MIN-SVI)

   *  Description: Selects the path with lowest Severe Violated Interval
      (SVI) occurrence, minimizing severe degradation events.

   The objective function is formulated using the following terminology:

   *  A network comprises a set of N links {Li, (i=1...N)}.

   *  A path P is a list of K links {Lpi,(i=1...K)}.

Contreras, et al.        Expires 7 January 2027                [Page 10]
Internet-Draft                PAM-based PCE                    July 2026

   *  Severe Violated Interval on link L is denoted SVI(L)

   *  The Severe Violated Intervals of a path P is denoted SVI(P), where
      SVI(P) = Max {SVI(Lpi), (i=1...K)}.

   *  The Min-SVI OF is to find a path P such that ( Max { SVI(Lpi),
      i=1...K } ) is minimized.

4.4.  Interaction with Path Constraints

   A PCC MAY request PAM information to be treated as a mandatory
   constraint or as a desirable optimization criterion.  Thus, when PAM
   information is expressed as a mandatory constraint, candidate paths
   not satisfying the requested PAM requirements SHALL be excluded from
   the solution set.  On the other hand, when PAM information is
   expressed as a desirable criterion, candidate paths not satisfying
   the requested PAM requirements MAY still be considered by the PCE.
   In such cases, PAM objective functions can be used to identify the
   most suitable path among the available alternatives.

   Potential mechanisms used to compare statistical distributions and
   determine the preferred candidate path are outside the scope of this
   document.

5.  PAM report TLV

   The PAM Report TLV is used by a PCE to report the statistical
   characterization associated with a computed path.  This TLV allows a
   PCC to understand the Precision Availability Metrics (PAM)
   information that has been used during path computation and path
   selection.  Such information can be particularly useful when the
   requested PAM profile has been expressed as a desirable objective
   rather than as a mandatory constraint.The PAM Report TLV MAY be
   included in PCRep messages and other PCEP messages carrying path
   computation results.

   The PAM Report TLV contains the statistical characterization
   associated with the selected path and MAY include:

   *  The performance metric being described (e.g., one-way delay,
      round-trip delay, jitter, packet loss).

   *  The observation interval used for deriving the statistical
      characterization.

   *  The kind of statistical distribution associated with the selected
      path.

Contreras, et al.        Expires 7 January 2027                [Page 11]
Internet-Draft                PAM-based PCE                    July 2026

   *  The distribution representation, including the set of bins or
      intervals used to characterize the metric values.  The encoding of
      specific statistical distributions and the representation of bins
      are outside the scope of this document.

   *  The corresponding PAM indicators, such as Violated Intervals (VI)
      and Severe Violated Intervals (SVI), as defined in [RFC9544].

   When the requested PAM profile cannot be satisfied exactly, the PAM
   Report TLV provides visibility about the statistical characteristics
   of the path actually selected by the PCE.  Moreover, when PAM
   information is requested as a desirable optimization criterion, the
   PCE MAY select a path whose statistical characterization differs from
   the one requested by the PCC.  In such case, the PCE SHOULD include a
   PAM Report TLV allowing the PCC to know (and evaluate) the
   characteristics of the selected path.

6.  Security and operational considerations

6.1.  Security considerations

   Same security and operational considerations as described in
   [RFC5440] apply also in this document.

   Other security considerations will be addressed in future versions of
   the document.

6.2.  Operational considerations

   The work with precision metrics can impose stringent requirements in
   terms of collection, processing and assessment of metrics of
   interest.  Such capabilities are expected to be supported by external
   systems, such as the TED, with the role of the PCE being limited to
   the work with processed information (e.g., histograms) so to assess
   that the precision metric used as constraint is compliant with the
   expectation of the PCC.  Such external supportive systems are out of
   scope of this document.

7.  IANA Considerations

7.1.  New PCE Objective Function Codes

   IANA is requested to assign the following values in the "Objective
   Function" subregistry within the "Path Computation Element Protocol
   (PCEP) Numbers" registry:

Contreras, et al.        Expires 7 January 2027                [Page 12]
Internet-Draft                PAM-based PCE                    July 2026

  +-----------+--------------------------------------------+--------------+
  |Code Point | Name                                       | Reference    |
  +-----------+--------------------------------------------+--------------+
  | TBD1      | PAM Compliance (PAM-COMP)                  | This document|
  | TBD2      | Minimum Violated Intervals (MIN-VI)        | This document|
  | TBD3      | Minimum Severe Violated Intervals (MIN-SVI)| This document|
  +--------------+-----------------------------------------+--------------+

7.2.  PCEP TLV

   IANA is requested to assign the following values in the "PCEP TLV
   Type Indicators" subregistry within the "Path Computation Element
   Protocol (PCEP) Numbers" registry:

     +----------+--------------------------------+--------------+
     |TLV Type  | Name                           | Reference    |
     +----------+--------------------------------+--------------+
     | TBD4     | PAM Report                     | This document|
     +----------+--------------------------------+--------------+

8.  Informative References

   [I-D.ietf-nmop-network-incident-yang]
              Hu, T., Contreras, L. M., Wu, Q., Davis, N., and C. Feng,
              "A YANG Data Model for Network Incident Management", Work
              in Progress, Internet-Draft, draft-ietf-nmop-network-
              incident-yang-10, 6 July 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-nmop-
              network-incident-yang-10>.

   [IANA_METRIC_Object]
              "METRIC Object T Field", n.d.,
              <https://www.iana.org/assignments/pcep/pcep.xhtml#metric-
              object-ni-field>.

   [IEEE.754.2019]
              "754-2019 - IEEE Standard for Floating-Point Arithmetic",
              22 July 2019,
              <https://ieeexplore.ieee.org/document/8766229>.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/info/rfc2119>.

Contreras, et al.        Expires 7 January 2027                [Page 13]
Internet-Draft                PAM-based PCE                    July 2026

   [RFC4655]  Farrel, A., Vasseur, J.-P., and J. Ash, "A Path
              Computation Element (PCE)-Based Architecture", RFC 4655,
              DOI 10.17487/RFC4655, August 2006,
              <https://www.rfc-editor.org/info/rfc4655>.

   [RFC5440]  Vasseur, JP., Ed. and JL. Le Roux, Ed., "Path Computation
              Element (PCE) Communication Protocol (PCEP)", RFC 5440,
              DOI 10.17487/RFC5440, March 2009,
              <https://www.rfc-editor.org/info/rfc5440>.

   [RFC5541]  Le Roux, JL., Vasseur, JP., and Y. Lee, "Encoding of
              Objective Functions in the Path Computation Element
              Communication Protocol (PCEP)", RFC 5541,
              DOI 10.17487/RFC5541, June 2009,
              <https://www.rfc-editor.org/info/rfc5541>.

   [RFC8233]  Dhody, D., Wu, Q., Manral, V., Ali, Z., and K. Kumaki,
              "Extensions to the Path Computation Element Communication
              Protocol (PCEP) to Compute Service-Aware Label Switched
              Paths (LSPs)", RFC 8233, DOI 10.17487/RFC8233, September
              2017, <https://www.rfc-editor.org/info/rfc8233>.

   [RFC8578]  Grossman, E., Ed., "Deterministic Networking Use Cases",
              RFC 8578, DOI 10.17487/RFC8578, May 2019,
              <https://www.rfc-editor.org/info/rfc8578>.

   [RFC8655]  Finn, N., Thubert, P., Varga, B., and J. Farkas,
              "Deterministic Networking Architecture", RFC 8655,
              DOI 10.17487/RFC8655, October 2019,
              <https://www.rfc-editor.org/info/rfc8655>.

   [RFC9543]  Farrel, A., Ed., Drake, J., Ed., Rokui, R., Homma, S.,
              Makhijani, K., Contreras, L., and J. Tantsura, "A
              Framework for Network Slices in Networks Built from IETF
              Technologies", RFC 9543, DOI 10.17487/RFC9543, March 2024,
              <https://www.rfc-editor.org/info/rfc9543>.

   [RFC9544]  Mirsky, G., Halpern, J., Min, X., Clemm, A., Strassner,
              J., and J. François, "Precision Availability Metrics
              (PAMs) for Services Governed by Service Level Objectives
              (SLOs)", RFC 9544, DOI 10.17487/RFC9544, March 2024,
              <https://www.rfc-editor.org/info/rfc9544>.

Contreras, et al.        Expires 7 January 2027                [Page 14]
Internet-Draft                PAM-based PCE                    July 2026

Appendix A.  Path Histogram Composition

   In order to obtain the statistical distribution of a metric over a
   complete path from the corresponding distributions of its constituent
   segments (e.g., hops) it is necessary to consider the class of the
   metric under evaluation, i.e., if the metric is additive,
   multiplicative, or maximal/minimal.

A.1.  Additive Metrics

   Additive metrics are those that sum along the path, such as delay or
   IGP cost [RFC4655], [RFC8233].  To generate a path histogram from
   segment histograms, the total path value can be obtained by summing
   the individual segment values along a period, and then forming the
   histogram.

   Alternatively, considering that a histogram is divided into discrete
   bins representing value ranges, it is possible to perform a bin-by-
   bin summation.  The histogram for the path is then obtained by
   summing the bin values across the segments.

A.2.  Multiplicative Metrics

   Multiplicative metrics, for example link availability or success
   probability [RFC8233], combine along a path by multiplying segment
   (e.g., per hop) values.  The path histogram can be obtained by
   combining the segment values and computing the product for each
   combination.

   Alternatively, logarithmic transformation can be applied to convert
   multiplicative aggregation into additive form, enabling reuse of
   additive histogram composition techniques.  In this method, the
   values of each histogram bin are transformed by taking the logarithm,
   effectively converting multiplication into addition.  The histograms
   can then be combined by summing the log-transformed bin values across
   segments, using the values of each bin per segment to calculate the
   resulting distribution.  After aggregating the histograms in the log
   domain, the path histogram can be transformed back to the original
   metric domain by applying the exponential function, yielding the
   final probabilities for the multiplicative path values.

A.3.  Maximization / Minimization Metrics (Bottleneck Metrics)

   Bottleneck metrics are defined by taking the maximum or minimum value
   along the path, such as bandwidth, MTU, etc [RFC4655].  To construct
   a path histogram, the values of each segment are considered to build
   the cumulative distribution function (CDF) of the path.

Contreras, et al.        Expires 7 January 2027                [Page 15]
Internet-Draft                PAM-based PCE                    July 2026

Acknowledgements

   The authors thank Dhruv Dhody, Rakesh Gandhi, Ruediger Geib, Amal
   Karboubi and Greg Mirsky for the comments received that helped to
   improve the document.

   This work has been partially funded by the European Commission
   Horizon Europe SNS JU PREDICT-6G project (GA 101095890), and the
   Spanish Ministry of Economic Affairs and Digital Transformation and
   the European Union NextGenerationEU UNICO 5G I+D "Towards a smart and
   efficient telecom infrastructure meeting current and future industry
   needs" (TIMING) project (TSI-063000-2021-145, -148, -149).

Authors' Addresses

   Luis M. Contreras
   Telefonica
   Ronda de la Comunicacion, s/n
   28050 Madrid
   Spain
   Email: luismiguel.contrerasmurillo@telefonica.com
   URI:   http://lmcontreras.com

   Fernando Agraz
   Universitat Politecnica de Catalunya
   08034 Barcelona
   Spain
   Email: fernando.agraz@upc.edu

   Salvatore Spadaro
   Universitat Politecnica de Catalunya
   08034 Barcelona
   Spain
   Email: salvatore.spadaro@upc.edu

   Quan Xiong
   ZTE Corporation
   China
   Email: xiong.quan@zte.com.cn

Contreras, et al.        Expires 7 January 2027                [Page 16]