Experiment: Network Anomaly Lifecycle
draft-netana-nmop-network-anomaly-lifecycle-01

NMOP                                                        V. Riccobene
Internet-Draft                                                A. Roberto
Intended status: Experimental                                     Huawei
Expires: 16 September 2024                                       T. Graf
                                                                   W. Du
                                                                Swisscom
                                                           A. Huang Feng
                                                               INSA-Lyon
                                                           15 March 2024

                 Experiment: Network Anomaly Lifecycle
             draft-netana-nmop-network-anomaly-lifecycle-01

Abstract

   Accurately detect network anomalies is very challenging for network
   operators in production networks.  Good results require a lot of
   expertise and knowledge around both the implied network technologies
   and the specific service provided to consumers, apart from a proper
   monitoring infrastructure.  In order to facilitate the detection of
   network anomalies, novel techniques are being introduced, including
   AI-based ones, with the promise of improving scalability and the hope
   to keep a high detection accuracy.  To guarantee acceptable results,
   the process needs to be properly designed, adopting well-defined
   stages to accurately collect evidence of anomalies, validate their
   relevancy and improve the detection systems over time.

   This document describes the lifecycle process to iteratively improve
   network anomaly detection accurately.  Three key stages are proposed,
   along with a YANG model specifying the required metadata for the
   network anomaly detection covering the different stages of the
   lifecycle.

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."

Riccobene, et al.       Expires 16 September 2024               [Page 1]
Internet-Draft          network-anomaly-lifecycle             March 2024

   This Internet-Draft will expire on 16 September 2024.

Copyright Notice

   Copyright (c) 2024 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.
   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.  Status of this document . . . . . . . . . . . . . . . . . . .   2
   2.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
   3.  Terminology . . . . . . . . . . . . . . . . . . . . . . . . .   4
   4.  Defining Desired States . . . . . . . . . . . . . . . . . . .   5
   5.  Lifecycle of a Network Anomaly  . . . . . . . . . . . . . . .   6
     5.1.  Network Anomaly Detection . . . . . . . . . . . . . . . .   7
     5.2.  Network Anomaly Validation  . . . . . . . . . . . . . . .   8
     5.3.  Network Anomaly Refinement  . . . . . . . . . . . . . . .   8
   6.  Network Anomaly State Machine . . . . . . . . . . . . . . . .   8
     6.1.  Overview of the Model for the Network Anomaly Metadata  .   9
   7.  Implementation status . . . . . . . . . . . . . . . . . . . .  14
     7.1.  Antagonist  . . . . . . . . . . . . . . . . . . . . . . .  14
   8.  Security Considerations . . . . . . . . . . . . . . . . . . .  14
   9.  Acknowledgements  . . . . . . . . . . . . . . . . . . . . . .  14
   10. Normative References  . . . . . . . . . . . . . . . . . . . .  14
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  16

1.  Status of this document

   This document is experimental.  The main goal of this document is to
   propose an iterative lifecycle process to network anomaly detection
   by proposing a data model for metadata to be addressed at different
   lifecycle stages.

   The experiment consists of verifying whether the approach is usable
   in real use case scenarios to support proper refinement and
   adjustments of network anomaly detection algorithms.  The experiments
   can be deemed successful if validated at least with an open-source
   implementation sucessfully applied in real production networks.

Riccobene, et al.       Expires 16 September 2024               [Page 2]
Internet-Draft          network-anomaly-lifecycle             March 2024

2.  Introduction

   In [Ahf23] network anomalies are defined as "Whatever would let an
   operator frown and investigate when looking at the collected
   forwarding plane, control plane and management plane network data
   relative to a customer".

   In [I-D.netana-nmop-network-anomaly-semantics] a semantic for the
   annotation of network anomalies has been defined in order to support
   the exchange of related metadata between different actors,
   formalizing a semantically consistent representation of the behaviors
   worth investigating.  In the same document, symptoms are defined as
   the essential piece of information to analyze network anomalies and
   incidents.

   The intention is to enable operators detecting network incidents
   timely.  A network incident can be defined as "An event that has a
   negative effect that is not as required/desired" (see
   [I-D.davis-nmop-incident-terminology]), or even more broadly, as "An
   unexpected interruption of a network service, degradation of network
   service quality, or sub-health of a network service" [TMF724A].

   With all this in mind, this document starts from the assumption that
   it is still remarkably difficult to gain a full understanding and a
   complete perspective of "if" and "how" the network is deviating from
   the desired state: on the one side, symptoms are not necessarily a
   guarantee of an incident happening (false positives), on the other
   side, the lack of symptom is not a guarantee of the absence of an
   incident (false negative).  The concept of network anomaly in this
   document plays the role of a bridge between symptoms and incident: a
   network anomaly is defined as a collection of symptoms, but without
   the guarantee that the observed symptoms are impacting existing
   services.  This opens up to the necessity of further validating the
   network anomalies to understand if the detected symptoms are actually
   impacting services.  This requires different actors (both human and
   algorithmic) to jump in during the process and refine their
   understanding across the network anomaly lifecycle.

Riccobene, et al.       Expires 16 September 2024               [Page 3]
Internet-Draft          network-anomaly-lifecycle             March 2024

   Performing network anomaly detection is a process that requires a
   continuous learning and continuous improvement.  Network anomalies
   are detected by collecting and understanding symptoms, then validated
   by confirming that there actually were service impacting and
   eventually need to be further analyzed by performing postmortem
   analysis to identify any potential adjustment to improve the
   detection capability.  Each of these stages is an opportunity to
   learn and refine the process, and since these stages might also be
   provided by different parties and/or products, this document
   contributes a formal structure to capture and exchange symptom
   information across the lifecycle.

3.  Terminology

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
   "OPTIONAL" in this document are to be interpreted as described in BCP
   14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

   This document makes use of the terms defined in
   [I-D.davis-nmop-incident-terminology].

   *  State

   *  Incident

   *  Event

   *  Alarm

   The following terms are used as defined in [RFC9417].

   *  Symptom

   *  Metric

   *  Intent

   The following terms are defined in this document.

   *  Author: Is a human or an algorithm which produces metadata by
      describing anomalies with symptoms.

   *  False Positive: Is a detected anomaly which has been identified
      during the postmortem to be not anomalous.

Riccobene, et al.       Expires 16 September 2024               [Page 4]
Internet-Draft          network-anomaly-lifecycle             March 2024

   *  False Negative: Is anomalous but has been not been identified by
      the anomaly detection system.

4.  Defining Desired States

   The above definitions of network incident provide the scope for what
   to be looking for when detecting network anomalies.  Concepts like
   "desirable state" and "required state" are introduced.  This poses
   the attention on a significant problem that network operators have to
   face: the definition of what is to be considered "desirable" or
   "undesirable".  It is not always easy to detect if a network is
   operating in an undesired state at a given point in time.  To
   approach this, network operators can rely on different methodologies,
   more or less deterministic and more or less sensitive: on the one
   side, the definition of intents (including Service Level Objectives
   and Service Level Agreements) which approaches the problem top-down;
   on the other side, the definition of symptoms, by mean of solutions
   like SAIN [RFC9417], [RFC9418] and Daisy [Ahf23], which approaches
   the problem bottom-up.  At the center of these approaches, there are
   the so-called symptoms, defined as reasons explaining what is not
   working as expected in the network, sometimes also providing hints
   towards issues and their causes.

   One of the more deterministic approaches is to rely on symptoms based
   on measurable service-based KPIs, for example, by using Service Level
   Indicators, Objectives and Agreements:

   Service Level Agreement (SLA)  An SLA is an agreement between parties
      that a service provider makes to its customers on the behavior of
      the provided service.  SLAs are a tool to define exactly what
      customers can expect out of the service provided to them.  In many
      cases, SLA breaches also come with contractual penalties.

   Service Level Objectives (SLOs)  An SLO is a threshold above which
      the service provider acts to prevent a breach of an SLA.  SLOs are
      a tool for service providers to know when they should start
      becoming concerned about a service not behaving as expected.  SLOs
      are rarely connected to penalties as they usually are internal
      metrics for the service providers.

   Service Level Indicators (SLIs)  An SLI is an observable metric that
      describes the state of a monitored subsystem.  SLIs are a tool to
      gain measurable visibility about the behavior of a subsystem in
      the network.  SLIs are usually the basis for SLOs, as the main
      difference between an SLI and SLO is that SLOs usually are defined
      as thresholds applied to SLIs.

Riccobene, et al.       Expires 16 September 2024               [Page 5]
Internet-Draft          network-anomaly-lifecycle             March 2024

   However, the definition of these KPIs turns out to be very
   challenging in some cases, as accurate KPIs could require
   computationally expensive techniques to be collected or substantial
   modifications to existing network protocols.

   Alternative methodologies rely on symptoms as the way to generate
   analytical data out of operational data.  For instance:

   SAIN  introduces the definition and exposure of symptoms as a
      mechanism for detecting those concerning behaviors in more
      deterministic ways.  Moreover, the concept of "impact score" has
      been introduced by SAIN, to indicate what is the expected degree
      of impact that a given symptom will have on the services relying
      on the related subservice to which the symptom is attached.

   Daisy  introduces the concept of concern score to indicate what is
      the degree of concern that a given symptom could cause a
      degradation for a service.

   In general, defining boundaries between desirable vs. undesirable in
   an accurate fashion requires continuous iterations and improvements
   coming from all the stages of the network anomaly detection
   lifecycle, by which network engineers can transfer what they learn
   through the process into new symptom definitions or refinements of
   the algorithms.

5.  Lifecycle of a Network Anomaly

   The lifecycle of a network anomaly can be articulated in three
   phases, structured as a loop: Detection, Validation, Refinement.

                                +-------------+
                    +--------> |  Detection  | ---------+
        Adjustments |          +-------------+          | Symptoms
                    |                                   |
                    |                                   v
            +------------+                       +------------+
            | Refinement |<--------------------- | Validation |
            +------------+        Incident       +------------+
                                Confirmation

              Figure 1: Anomaly Detection Refinement Lifecycle

   Each of these phases can either be performed by a network expert or
   an algorithm or complementing each other.

Riccobene, et al.       Expires 16 September 2024               [Page 6]
Internet-Draft          network-anomaly-lifecycle             March 2024

   The network anomaly metadata is generated by an author, which can be
   either a human expert or an algorithm.  The author can produce the
   metadata for a network anomaly, for each stage of the cycle and even
   multiple versions for the same stage.  In each version of the network
   anomaly metadata, the author indicates the list of symptoms that are
   part of the network anomaly taken into account.  The iterative
   process is about the identification of the right set of symptoms.

5.1.  Network Anomaly Detection

   The Network Anomaly Detection stage is about the continuous
   monitoring of the network through Network Telemetry [RFC9232] and the
   identification of symptoms.  One of the main requirements that
   operator have on network anomaly detection systems is the high
   accuracy.  This means having a small number of false negatives,
   symptoms causing service impact are not missed, and false positives,
   symptoms that are actually innocuous are not picked up.

   As the detection stage is becoming more and more automated for
   production networks, the identified symptoms might point towards
   three potential kinds of behaviors:

   i. those that are surely corresponding to an impact on services,
   (e.g. the breach of an SLO),

   ii. those that will cause problems in the future (e.g. rising trends
   on a timeseries metric hitting towards saturation),

   iii. those or which the impact to services cannot be confirmed (e.g.
   sudden increase/decrease of timeseries metrics, anomalous amounts of
   log entries, etc.).

   The first category requires immediate intervention (a.k.a. the
   incident is "confirmed"), the second one provides pointers towards
   early signs of an incident potentially happening in the near future
   (a.k.a. the incident is "forecasted"), and the third one requires
   some analysis to confirm if the detected symptom requires any
   attention or immediate intervention (a.k.a. the incident is
   "potential").  As part of the iterative improvement required in this
   stage, one that is very relevant is the gradual conversion of the
   third category into one of the first two, which would make the
   network anomaly detection system more deterministic.  The main
   objective is to reduce uncertainty around the raised alarms by
   refining the detection algorithms.  This can be achieved by either
   generating new symptom definitions, adjusting the weights of
   automated algorithms or other similar approaches.

Riccobene, et al.       Expires 16 September 2024               [Page 7]
Internet-Draft          network-anomaly-lifecycle             March 2024

5.2.  Network Anomaly Validation

   The key objective for the validation stage is clearly to decide if
   the detected symptoms are signaling a real incident (a.k.a. require
   immediate action) or if they are to be treated as false positives
   (a.k.a. suppressing the alarm).  For those symptoms surely having
   impact on services, 100% confidence on the fact that a network
   incident is happening can be assumed.  For the other two categories,
   "forecasted" and "potential", further analysis and validation is
   required.

5.3.  Network Anomaly Refinement

   After validation of an incident, the service provider has to perform
   troubleshooting and resolution of the incident.  Although the network
   might be back in a desired state at this point, network operators can
   perform detailed postmortem analysis of network incidents with the
   objective to identify useful adjustments to the prevention and
   detection mechanisms (for instance improving or extending the
   definition of SLIs and SLOs, refining concern/impact scores, etc.),
   and improving the accuracy of the validation stage (e.g. automating
   parts of the validation, implementing automated root cause analysis
   and automation for remediation actions).  In this stage of the
   lifecycle it is assumed that the incident is under analysis.

   After the adjustments are performed to the network anomaly detection
   methods, the cycle starts again, by "replaying" the network anomaly
   and checking if there is any measurable improvement in the ability to
   detect incidents by using the updated method.

6.  Network Anomaly State Machine

   From a network anomaly detection point of view a network incident is
   defined as a collection of interrelated symptoms.  From this
   perspective, an incident can be defined according to the following
   states (Figure 2).

Riccobene, et al.       Expires 16 September 2024               [Page 8]
Internet-Draft          network-anomaly-lifecycle             March 2024

                                             +---------+
                                             | Initial |-----------------+
                                             +---------+                 |
                                                  |                      |
                                            +-----+---------+            |
                                   +--------|---------------|------+     |
                                   | +------v-----+  +------v----+ |     |
                                   | |  Incident  |  |  Incident | |     |
                             +---->| | Forecasted |  | Potential | |     |
                             |     | +------------+  +-----------+ |     |
                             |     +--------|--Detection---|-------+     |
                             |              |              |             |
        +-------+            |              +------- ----- +             |
        | Final |            |                      |                    |
        +---^---+            |                      |                    |
            |                |                      |                    |
            |                |                      v                    |
            |                |     +-----------Validation------------+   |
+-----------------------+    |     |  +-----------+                  |   |
|           |           |    |     |  |  Network  |   +-----------+  |   |
|  +-----------------+  |    |     |  |  Anomaly  |   |  Incident |  |   |
|  | Network Anomaly |  |    |     |  | Discarded |   | Confirmed |<-|---+
|  |     Adjusted    |-------+     |  +-----|-----+   +-----------+  |
|  +--------^--------+  |          +---------------------------------+
|           |           |                   |               |
|           |           |               +---v---+           |
|           |           |               | Final |           |
|           |           |               +-------+           |
| +---------|--------+  |                                   |
| | Network Anomaly  |  |                                   |
| |     Analyzed     |<-|-----------------------------------+
| +------------------+  |
+-------Refinement------+

               Figure 2: Network Anomaly State Machine

6.1.  Overview of the Model for the Network Anomaly Metadata

Riccobene, et al.       Expires 16 September 2024               [Page 9]
Internet-Draft          network-anomaly-lifecycle             March 2024

        module: ietf-network-anomaly-metadata
          +--rw network-anomalies
             +--rw network-anomaly* [id author-name version state]
                +--rw id             yang:uuid
                +--rw description?   string
                +--rw author
                |  +--rw author-name     string
                |  +--rw author-type?    identityref
                |  +--rw algo-version?   uint8
                +--rw version        uint8
                +--rw state          identityref
                +--rw symptoms* [symptom_id]
                   +--rw symptom_id    yang:uuid

       Figure 3: YANG tree diagram for ietf-network-anomaly-metadata

   <CODE BEGINS> file "ietf-network-anomaly-metadata@2024-02-26.yang"
   module ietf-network-anomaly-metadata {
     yang-version 1.1;
     namespace "urn:ietf:params:xml:ns:yang:ietf-network-anomaly-metadata";
     prefix network_anomaly_metadata;

     import ietf-yang-types {
       prefix yang;
       reference "RFC 6991: Common YANG Data Types";
     }

     organization
       "IETF NMOP Working Group";
     contact
       "WG Web:   <https://datatracker.ietf.org/wg/nmop/>
        WG List:  <mailto:nmop@ietf.org>

        Authors:  Vincenzo Riccobene
                  <mailto:vincenzo.riccobene@huawei-partners.com>
                  Antonio Roberto
                  <mailto:antonio.roberto@huawei.com>
                  Thomas Graf
                  <mailto:thomas.graf@swisscom.com>
                  Wanting Du
                  <mailto:wanting.du@swisscom.com>
                  Alex Huang Feng
                  <mailto:alex.huang-feng@insa-lyon.fr>";
     description
       "This module defines objects for the description of network anomalies.
         Network anomalies are a collection of symptoms observed on
         the network nodes.

Riccobene, et al.       Expires 16 September 2024              [Page 10]
Internet-Draft          network-anomaly-lifecycle             March 2024

         Copyright (c) 2024 IETF Trust and the persons identified as
         authors of the code.  All rights reserved.

         Redistribution and use in source and binary forms, with or
         without modification, is permitted pursuant to, and subject
         to the license terms contained in, the Revised BSD License
         set forth in Section 4.c of the IETF Trust's Legal Provisions
         Relating to IETF Documents
         (https://trustee.ietf.org/license-info).

         This version of this YANG module is part of RFC XXXX; see the RFC
         itself for full legal notices.";

     revision 2024-02-26 {
       description
         "Initial version";
       reference
         "RFCXXXX: Experiment: Network Anomaly Postmortem Lifecycle";
     }

     identity author-type {
       description
         "Type of the author of the network anomaly metadata";
     }

     identity user {
       base author-type;
       description
         "A real user (person) generated the network anomaly metadata";
     }

     identity algorithm {
       base author-type;
       description
         "An algorithm generated the network anomaly metadata";
     }

     identity network-anomaly-state {
       description
         "Base identity for representing the state of the network anomaly";
     }
     identity incident-forecasted {
       base network-anomaly-state;
       description
         "An incident has been forecasted, as it is expected that
         the indicated list of symptoms will impact a service
         in the near future";
     }

Riccobene, et al.       Expires 16 September 2024              [Page 11]
Internet-Draft          network-anomaly-lifecycle             March 2024

     identity incident-potential {
       base network-anomaly-state;
       description
         "An incident has been detected with a confidence
         lower than 100%. In order to confirm that this set of
         symptoms are generating service impact, it requires further
         validation";
     }
     identity incident-confirmed {
       base network-anomaly-state;
       description
         "After validation, the incident has been confirmed";
     }
     identity discarded {
       base network-anomaly-state;
       description
         "After validation, the network anomaly has been
         discarded, as there is no evindence that it is causing an
         incident";
     }
     identity analysed {
       base network-anomaly-state;
       description
         "The anomaly detection went through analysis to identify
         potential ways to further improve the detection process in
         for future anomalies";
     }
     identity adjusted {
       base network-anomaly-state;
       description
         "The network anomaly has been solved and analysed.
         No further action is required.";
     }

     container network-anomalies {
       description "Container having the network anomalies";
       list network-anomaly {
         key "id author-name version state";
         description "A network anomaly identified by an id, author-name, version
           and state.";
         leaf id {
           type yang:uuid;
           description
               "Unique ID of the network network anomaly";
         }
         leaf description {
           type string;
           description

Riccobene, et al.       Expires 16 September 2024              [Page 12]
Internet-Draft          network-anomaly-lifecycle             March 2024

             "Textual description of the network anomaly";
         }
         container author {
           description "Container defining the type of the author and the
             version of the algorithm if it is an algorithm who reported the anomaly.";
           leaf author-name {
             type string;
             description "Name of the user (person) or of the
               algorithm that generated the network anomaly metadata";
           }
           leaf author-type {
             type identityref {
                 base author-type;
             }
             description "The type of author who reported the anomaly.";
           }
           leaf algo-version {
             type uint8;
             description "Version of the algorithm used to
             produce the netowrk anomaly metadata.  This is
             used only if the author type is an algorithm";
           }
         }
         leaf version {
           type uint8;
           description
             "Version of the incident metadata object.
             It allows multiple versions of the metadata to be
             generated in order to support the definition of
             multiple incindent objects from the same source to
             facilitate improvements overtime";
         }
         leaf state {
           type identityref {
             base network-anomaly-state;
           }
           mandatory true;
           description "State of the anomaly.";
         }
         list symptoms {
           key "symptom_id";
           description "List of symptoms identified by the symptom_id.";
           leaf symptom_id {
             type yang:uuid;
             description "UUID of the symptom that is part of this incident";
           }
         }
       }

Riccobene, et al.       Expires 16 September 2024              [Page 13]
Internet-Draft          network-anomaly-lifecycle             March 2024

     }
   }
   <CODE ENDS>

          Figure 4: YANG module for ietf-network-anomaly-metadata

7.  Implementation status

   This section provides pointers to existing open source
   implementations of this draft.  Note to the RFC-editor: Please remove
   this before publishing.

7.1.  Antagonist

   A tool called Antagonist has been implemented during the IETF 119
   Hackathon, in order to validate the application of the YANG models
   defined in this draft.  Antagonist provides visual support for two
   important use cases in the scope of this document:

   *  the generation of a ground truth in relation to symptoms and
      incidents in timeseries data

   *  the visual validation of results produced by automated network
      anomaly detection tools.

   The open source code can be found here: [Antagonist]

8.  Security Considerations

   The security considerations will have to be updated according to
   "https://wiki.ietf.org/group/ops/yang-security-guidelines".

9.  Acknowledgements

   The authors would like to thank xxx for their review and valuable
   comments.

10.  Normative References

   [Ahf23]    Huang Feng, A., "Daisy: Practical Anomaly Detection in
              large BGP/MPLS and BGP/SRv6 VPN Networks", IETF 117,
              Applied Networking Research Workshop,
              DOI 10.1145/3606464.3606470, July 2023,
              <https://anrw23.hotcrp.com/doc/anrw23-paper8.pdf>.

Riccobene, et al.       Expires 16 September 2024              [Page 14]
Internet-Draft          network-anomaly-lifecycle             March 2024

   [Antagonist]
              Riccobene, V., Roberto, A., Du, W., Graf, T., and H. Huang
              Feng, "Antagonist: Anomaly tagging on historical data",
              <https://github.com/vriccobene/antagonist>.

   [I-D.davis-nmop-incident-terminology]
              Davis, N. and A. Farrel, "Some Key Terms for Incident
              Management", Work in Progress, Internet-Draft, draft-
              davis-nmop-incident-terminology-00, 18 January 2024,
              <https://datatracker.ietf.org/doc/html/draft-davis-nmop-
              incident-terminology-00>.

   [I-D.netana-nmop-network-anomaly-semantics]
              Graf, T., Du, W., Feng, A. H., Riccobene, V., and A.
              Roberto, "Semantic Metadata Annotation for Network Anomaly
              Detection", Work in Progress, Internet-Draft, draft-
              netana-nmop-network-anomaly-semantics-01, 15 March 2024,
              <https://datatracker.ietf.org/api/v1/doc/document/draft-
              netana-nmop-network-anomaly-semantics/>.

   [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>.

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/info/rfc8174>.

   [RFC8340]  Bjorklund, M. and L. Berger, Ed., "YANG Tree Diagrams",
              BCP 215, RFC 8340, DOI 10.17487/RFC8340, March 2018,
              <https://www.rfc-editor.org/info/rfc8340>.

   [RFC9232]  Song, H., Qin, F., Martinez-Julia, P., Ciavaglia, L., and
              A. Wang, "Network Telemetry Framework", RFC 9232,
              DOI 10.17487/RFC9232, May 2022,
              <https://www.rfc-editor.org/info/rfc9232>.

   [RFC9417]  Claise, B., Quilbeuf, J., Lopez, D., Voyer, D., and T.
              Arumugam, "Service Assurance for Intent-Based Networking
              Architecture", RFC 9417, DOI 10.17487/RFC9417, July 2023,
              <https://www.rfc-editor.org/info/rfc9417>.

   [RFC9418]  Claise, B., Quilbeuf, J., Lucente, P., Fasano, P., and T.
              Arumugam, "A YANG Data Model for Service Assurance",
              RFC 9418, DOI 10.17487/RFC9418, July 2023,
              <https://www.rfc-editor.org/info/rfc9418>.

Riccobene, et al.       Expires 16 September 2024              [Page 15]
Internet-Draft          network-anomaly-lifecycle             March 2024

   [TMF724A]  TMF, "Incident Management API Profile v1.0.0", 3 April
              2023, <https://www.tmforum.org/resources/standard/tmf724a-
              incident-management-api-profile-v1-0-0/>.

Authors' Addresses

   Vincenzo Riccobene
   Huawei
   Dublin
   Ireland
   Email: vincenzo.riccobene@huawei-partners.com

   Antonio Roberto
   Huawei
   Dublin
   Ireland
   Email: antonio.roberto@huawei.com

   Thomas Graf
   Swisscom
   Binzring 17
   CH-8045 Zurich
   Switzerland
   Email: thomas.graf@swisscom.com

   Wanting Du
   Swisscom
   Binzring 17
   CH-8045 Zurich
   Switzerland
   Email: wanting.du@swisscom.com

   Alex Huang Feng
   INSA-Lyon
   Lyon
   France
   Email: alex.huang-feng@insa-lyon.fr

Riccobene, et al.       Expires 16 September 2024              [Page 16]