ETSI GR NFV-EVE 027 V6.1.1 (2026-04)

Network Functions Virtualisation (NFV) Release 6; Evolution and Ecosystem; Report on Model-as-a-Service (MaaS) in NFV

DGR/NFV-EVE027

General Information

• Status: Not Published
• Technical Committee: NFV EVE - Evolution and Ecosystem
• Current Stage: 12 - Citation in the OJ (auto-insert)
• Due Date: 03-Apr-2026
• Completion Date: 08-Apr-2026

Standards Content (Sample)

GROUP REPORT
Network Functions Virtualisation (NFV) Release 6;
Evolution and Ecosystem;
Report on Model-as-a-Service (MaaS) in NFV

Disclaimer
The present document has been produced and approved by the Network Functions Virtualisation (NFV) ETSI Industry
Specification Group (ISG) and represents the views of those members who participated in this ISG.
It does not necessarily represent the views of the entire ETSI membership.

2 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)

Reference
DGR/NFV-EVE027
Keywords
AI, MaaS, management, NFV
ETSI
650 Route des Lucioles
F-06921 Sophia Antipolis Cedex - FRANCE

Tel.: [phone]  Fax: [phone]

Siret N° 348 623 562 00017 - APE 7112B
Association à but non lucratif enregistrée à la
Sous-Préfecture de Grasse (06) N° w061004871

Important notice
The present document can be downloaded from the
ETSI Search & Browse Standards application.
The present document may be made available in electronic versions and/or in print. The content of any electronic and/or
print versions of the present document shall not be modified without the prior written authorization of ETSI. In case of any
existing or perceived difference in contents between such versions and/or in print, the prevailing version of an ETSI
deliverable is the one made publicly available in PDF format on ETSI deliver repository.
Users should be aware that the present document may be revised or have its status changed,
this information is available in the Milestones listing.
If you find errors in the present document, please send your comments to
the relevant service listed under Committee Support Staff.
If you find a security vulnerability in the present document, please report it through our
Coordinated Vulnerability Disclosure (CVD) program.
Notice of disclaimer & limitation of liability
The information provided in the present deliverable is directed solely to professionals who have the appropriate degree of
experience to understand and interpret its content in accordance with generally accepted engineering or
other professional standard and applicable regulations.
No recommendation as to products and services or vendors is made or should be implied.
No representation or warranty is made that this deliverable is technically accurate or sufficient or conforms to any law
and/or governmental rule and/or regulation and further, no representation or warranty is made of merchantability or fitness
for any particular purpose or against infringement of intellectual property rights.
In no event shall ETSI be held liable for loss of profits or any other incidental or consequential damages.

Any software contained in this deliverable is provided "AS IS" with no warranties, express or implied, including but not
limited to, the warranties of merchantability, fitness for a particular purpose and non-infringement of intellectual property
rights and ETSI shall not be held liable in any event for any damages whatsoever (including, without limitation, damages
for loss of profits, business interruption, loss of information, or any other pecuniary loss) arising out of or related to the use
of or inability to use the software.
Copyright Notification
No part may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and
microfilm except as authorized by written permission of ETSI.
The content of the PDF version shall not be modified without the written authorization of ETSI.
The copyright and the foregoing restriction extend to reproduction in all media.

© ETSI 2026.
All rights reserved.
ETSI
3 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
Contents
Intellectual Property Rights . 5
Foreword . 5
Modal verbs terminology . 5
1 Scope . 6
2 References . 6
2.1 Normative references . 6
2.2 Informative references . 6
3 Definition of terms, symbols and abbreviations . 7
3.1 Terms . 7
3.2 Symbols . 7
3.3 Abbreviations . 7
4 Introduction and overview . 7
4.1 Background information . 7
4.1.1 Introduction to large model technology . 7
4.1.2 Challenges in large model technology . 8
4.1.3 MaaS enables seamless AI deployment and utilization . 8
4.1.4 Relevant work in other SDOs . 9
5 Use cases . 9
5.1 Overview . 9
5.2 Use case #1: Root cause identification of Telco cloud failures . 9
5.2.1 Introduction. 9
5.2.2 Actors and roles . 10
5.2.3 Trigger . 10
5.2.4 Pre-conditions . 10
5.2.5 Post-conditions . 10
5.2.6 Flow description . 11
5.3 Use case #2: Query of Telco cloud operational metrics . 11
5.3.1 Introduction. 11
5.3.2 Actors and roles . 11
5.3.3 Trigger . 12
5.3.4 Pre-conditions . 12
5.3.5 Post-conditions . 12
5.3.6 Flow description . 12
5.4 Use case #3: Smart deployment plan generation for Telco cloud . 13
5.4.1 Introduction. 13
5.4.2 Actors and roles . 13
5.4.3 Trigger . 14
5.4.4 Pre-conditions . 14
5.4.5 Post-conditions . 14
5.4.6 Flow description . 14
5.5 Use case #4: Optimizing intent negotiation with large models . 15
5.5.1 Introduction. 15
5.5.2 Actors and roles . 15
5.5.3 Trigger . 16
5.5.4 Pre-conditions . 16
5.5.5 Post-conditions . 16
5.5.6 Flow description . 16
6 Key issue analysis . 17
6.1 Key issue on providing large models and large model applications for Telco cloud management

purposes . 17
6.2 Introducing and exposing large models as a service. 18
7 Framework and potential solutions . 18
ETSI
4 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
7.1 Introduction . 18
7.2 Potential solutions . 19
7.2.1 Solution #1: Integrating large models for telco cloud management . 19
7.2.1.1 Introduction . 19
7.2.1.2 Solution description . 19
7.2.1.3 Key issues address . 20
7.2.1.4 Gap analysis . 20
7.2.2 Solution #2: large model application deployment with separation of models and application logic

components . 21
7.2.2.1 Introduction . 21
7.2.2.2 Solution description . 21
7.2.2.3 Key issues address . 22
7.2.2.4 Gap analysis . 22
8 Recommendations . 23
8.1 Overview . 23
8.2 Recommendations related to the NFV architectural framework . 23
8.3 Recommendations related to interfaces and information model. 23
8.4 Recommendations related to NFV descriptors . 23
9 Conclusion . 24
History . 25

ETSI
5 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
Intellectual Property Rights
Essential patents
IPRs essential or potentially essential to normative deliverables may have been declared to ETSI. The declarations
pertaining to these essential IPRs, if any, are publicly available for ETSI members and non-members, and can be
found in ETSI SR 000 314: "Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to
ETSI in respect of ETSI standards", which is available from the ETSI Secretariat. Latest updates are available on the
ETSI IPR online database.
Pursuant to the ETSI Directives including the ETSI IPR Policy, no investigation regarding the essentiality of IPRs,
including IPR searches, has been carried out by ETSI. No guarantee can be given as to the existence of other IPRs not
referenced in ETSI SR 000 314 (or the updates on the ETSI Web server) which are, or may be, or may become,
essential to the present document.
Trademarks
The present document may include trademarks and/or tradenames which are asserted and/or registered by their owners.
ETSI claims no ownership of these except for any which are indicated as being the property of ETSI, and conveys no
right to use or reproduce any trademark and/or tradename. Mention of those trademarks in the present document does
not constitute an endorsement by ETSI of products, services or organizations associated with those trademarks.
DECT™, PLUGTESTS™, UMTS™ and the ETSI logo are trademarks of ETSI registered for the benefit of its
Members. 3GPP™, LTE™ and 5G™ logo are trademarks of ETSI registered for the benefit of its Members and of the
3GPP Organizational Partners. oneM2M™ logo is a trademark of ETSI registered for the benefit of its Members and of ®
the oneM2M Partners. GSM and the GSM logo are trademarks registered and owned by the GSM Association.
Foreword
This Group Report (GR) has been produced by ETSI Industry Specification Group (ISG) Network Functions
Virtualisation (NFV).
Modal verbs terminology
In the present document "should", "should not", "may", "need not", "will", "will not", "can" and "cannot" are to be
interpreted as described in clause 3.2 of the ETSI Drafting Rules (Verbal forms for the expression of provisions).
"must" and "must not" are NOT allowed in ETSI deliverables except when used in direct citation.

ETSI
6 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
1 Scope
The present document investigates Model-as-a-Service (MaaS) for AI-based applications in the context of telco cloud
management. It describes and analyses a set of relevant use cases, with a focus on the definition and role of MaaS
within the telco cloud management.
The present document also describes key issues, potential solutions, and where applicable, it also provides
recommendations for enhancements to the NFV architecture and its functionality. The proposed solutions are aiming to
provide further support for how MaaS can be introduced to the NFV framework and how MaaS can assist the
management of telco cloud services.
2 References
2.1 Normative references
Normative references are not applicable in the present document.
2.2 Informative references
References are either specific (identified by date of publication and/or edition number or version number) or
non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the
referenced document (including any amendments) applies.
NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee
their long term validity.
The following referenced documents may be useful in implementing an ETSI deliverable or add to the reader's
understanding, but are not required for conformance to the present document.
[i.1] ETSI GR NFV 003: "Network Functions Virtualisation (NFV); Terminology for Main Concepts in
NFV".
[i.2] ETSI GS NFV-IFA 047: "Network Functions Virtualisation (NFV) Release 5; Management and
Orchestration; Management data analytics Service Interface and Information Model
Specification".
[i.3] ETSI GS NFV-IFA 049: "Network Functions Virtualisation (NFV) Release 5; Architectural
Framework; VNF generic OAM functions and other PaaS Services specification".
[i.4] ETSI GS NFV-IFA 050: "Network Functions Virtualisation (NFV) Release 5; Management and
Orchestration; Intent Management Service Interface and Information Model Specification".
[i.5] ETSI GR NFV-IFA 054: "Network Functions Virtualisation (NFV) Release 6; Architecture;
Report on architectural support for NFV evolution".
[i.6] ETSI GR ENI 045 (V4.1.1): "Experiential Networked Intelligence (ENI); Research on Application
Scenarios of Network Large Language Models for Operation, Administration, Maintenance, and
Performance".
[i.7] ETSI GS ENI 050 (V4.5.1): "Experiential Networked Intelligence (ENI); Lifecycle management
for large model".
[i.8] 3GPP TR 28.858: "Study on Artificial Intelligence / Machine Learning (AI/ML) management
Phase 2 (Release 19)".
[i.9] ETSI TS 128 105: "5G; Management and orchestration; Artificial Intelligence/ Machine Learning
(AI/ML) management (3GPP TS 28.105 Release 19)".
[i.10] TMForum TMF780: "MaaS API Profile".
ETSI
7 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
[i.11] ETSI GR NFV-IFA 046: "Network Functions Virtualisation (NFV) Release 5; Architectural
Framework; Report on NFV support for virtualisation of RAN".
3 Definition of terms, symbols and abbreviations
3.1 Terms
For the purposes of the present document, the terms given in ETSI GR NFV 003 [i.1] and the following apply:
NOTE: A term defined in the present document takes precedence over the definition of the same term, if any, in
ETSI GR NFV 003 [i.1].
large model application: type of application that invokes large models according to specific requirements to
accomplish tasks within a particular application scenario
large-scale model: type of artificial neural network model with complex structure and large number of parameters,
enabling it to learn complex patterns and perform tasks with high accuracy
large-scale pre-trained model: type of a large-scale model that is already trained on massive datasets before being
adapted to specific tasks
3.2 Symbols
Void.
3.3 Abbreviations
For the purposes of the present document, the abbreviations given in ETSI GR NFV 003 [i.1] and the following apply:
AI Artificial Intelligence
LLD Low-Level Design
MaaS Model-as-a-Service
ML Machine Learning
RAG Retrieval-Augmented Generation
4 Introduction and overview
4.1 Background information
4.1.1 Introduction to large model technology
Currently, the integration of Artificial Intelligence (AI) technology to assist in telco cloud management has entered the
maturity phase. In the context of NFV AI technology can be utilized to enhance management within various domains,
including support for the management data analytics service defined in ETSI GS NFV-IFA 047 [i.2], the generic OAM
functions defined in ETSI GS NFV-IFA 049 [i.3], the intent management service defined in ETSI
GS NFV-IFA 050 [i.4], among others. With the rapid advancements in AI technology, particularly the ongoing
breakthroughs in reinforcement learning, large models, and content generation, the adoption of large models as the
foundation for downstream tasks (such as anomaly detection, intelligent maintenance, and intent recognition) has
emerged as a new paradigm in AI-based application domains.
The industry often follows a progression from foundational large models to industry-specific large models. Initially,
large models are trained to be applicable to multiple domains and tasks, equipping them with robust generalization
capabilities that enable extrapolation from one instance to another. Subsequently, industry-related data is used for pre-
training or fine-tuning to enhance performance and accuracy within that specific domain.
ETSI
8 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
Based on application scenarios and functionalities, large models can be categorized into several types: large language
models (for processing natural language text), vision large models (for processing images and videos), structured large
models (for processing structured data), and multimodal large models (for processing multi-modal data such as text,
images, audio, etc.). These models have the potential to empower various application scenarios within telco cloud
management.
Compared to traditional small models, large models are better equipped to handle the uncertainties and complexities of
the network domain, while small models still possess noteworthy advantages in tasks related to resource utilization,
interpretability, and stability. Therefore, it is necessary to rationally orchestrate the workflows within the network
domain, leveraging the collaboration between large and small models to achieve complementary advantages and
enhance the overall system's performance and efficiency.
In the realm of telco cloud management, leveraging the generalization capabilities of general-purpose large models and
incorporating exclusive domain-specific data from the telco cloud management field, capabilities such as intent
understanding, human-computer interaction, task decomposition, text generation, analysis, and reasoning can be
embedded into different application scenarios. Such scenarios include for example operation and maintenance
knowledge querying, data statistics analysis, network fault sensing, event delimitation and localization, and fault
handling closed loops.
Based on this foundation, large models can be used to gain a deep understanding of the complex dynamics of cloudified
network environments, providing robust support for real-time optimization of network resources, fault prediction, and
continuous improvement of service quality. This ultimately enhances the intelligence and automation capabilities of
telco cloud management processes significantly.
4.1.2 Challenges in large model technology
As large models continue to advance rapidly and application needs grow, the technical and economic costs have risen
sharply, bringing along complex challenges.
Firstly, deploying large models requires massive computational resources and data processing capabilities. With the
rapid expansion of model parameters, both training and subsequent model inference, which require significant
computational support, are leading to increasing costs.
Secondly, large models are technically complex regarding training, optimization, inference, and deployment. They have
stricter requirements by means of dataset structure and quality. Additionally, new techniques such as prompt
engineering (the process of structuring instructions that can be interpreted and understood by a large model) have
further raised the technical bar. This is due to the fact that prompt engineering requires a deep understanding of large
model capabilities and limitations, as well as the ability to craft instructions that effectively combine linguistic and
technical knowledge.
Furthermore, redundancy in model development is a notable issue, resulting not only in the waste of valuable resources
such as computational power, data storage, and development time, but also leading to a reduction in overall efficiency.
This occurs when multiple models are created that perform similar tasks, or when models are developed from scratch
without leveraging existing solutions or frameworks.
Lastly, the challenges of adapting large models to different scenarios and improving development efficiency urge for
attention. Due to their complexity and diversity, efficiently deploying large models in specific business contexts
remains a technical challenge to overcome.
NOTE: Distinctions between large models and small models in NFV for existing activities like in ETSI
GS NFV-IFA 047 [i.2], ETSI GS NFV-IFA 049 [i.3], and ETSI GS NFV-IFA 050 [i.4] will be explored
in future releases.
4.1.3 MaaS enables seamless AI deployment and utilization
MaaS (Model as a Service) encapsulates AI models and their associated capabilities into reusable services, enabling
users to swiftly build, deploy, monitor, and invoke models without the need to develop and maintain underlying
foundational capabilities. It offers a comprehensive suite of platform tools that streamline model training, tuning, and
deployment, empowering users to efficiently customize and bring models into operation. Additionally, MaaS in
principle integrates an extensive model library and dataset, thereby eliminating redundant development work.
ETSI
9 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
Furthermore, MaaS features robust application development capabilities, providing platforms or tools tailored to
specific scenarios. These enable users to rapidly construct AI applications. The MaaS services can be used to support a
range of model enhancement techniques, such as Retrieval-Augmented Generation (RAG), collaboration between large
and small models, delivering high-quality services in the form of intelligent agents tailored to different scenarios.
Intelligent agents in NFV can be considered autonomous software entities embedded within the NFV framework that
leverage AI/ML and data-driven reasoning to optimize, manage, and control virtualised network resources and services.
MaaS's capabilities encompass three key aspects: pure model inference services (providing direct access to trained
models for inference without involving model training or optimization), application-invoked model composite agent
services, and integrated management of application/model research, development, and operations. These comprehensive
services not only lower technical barriers and encourage model sharing but also enhance application adaptability.
Therefore, MaaS facilitates the widespread adoption and efficient utilization of AI models, driving transformation and
growth across various industries.
4.1.4 Relevant work in other SDOs
Many SDOs have already carried out research work related to large models and MaaS. In ETSI GR ENI 045 [i.6],
research was conducted on how to leverage large model technologies to assist in communication network operations
and management. In ETSI GS ENI 050 [i.7], research was conducted on how to encapsulate large models as a service
and manage the lifecycle of such services. In 3GPP TR 28.858 [i.8], research on requirements for generative AI
(involving large models) is also included, while ETSI TS 128 105 [i.9] defines the ML model lifecycle, ML model
lifecycle management capabilities and information model definitions for AI/ML management. In TMForum
TMF780 [i.10], APIs related to MaaS are defined.
® ®
The open-source organizations CAMARA under the Linux Foundation and ONAP (Open Network Automation ®
Platform) under Linux Foundation Networking have carried out development work on functions and requirements
related to MaaS. ®
NOTE: Linux is the registered trademark of Linus Torvalds in the U.S. and other countries.
The present document builds upon the existing work of the aforementioned standards and open-source organizations to
investigate how MaaS can be introduced to the NFV framework and how MaaS can assist the management of Telco
clouds.
5 Use cases
5.1 Overview
Packaging models could assist in various Telco cloud scenarios. For example the combination of large models, small
models, various tools, knowledge bases, and other components can be used to form diverse large model applications
supporting tasks like in telco cloud maintenance management and operation, among others.
In the context of maintenance management, large model services facilitate automated querying of performance metrics
and alarms within Telco clouds. Furthermore, by leveraging Telco cloud fault alarms and related performance data,
large models help pinpoint the root causes of faults.
Within operational scenarios, large model technology can assist in the deployment of Telco services. For instance,
during the assisted generation of LLD (Low-Level Design) documentation for Telco clouds, a large model specific to
Telco clouds provides guidance on configuring and selecting options for LLD documentation creation, thus enabling
efficient automation of the documentation process.
ETSI
10 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
5.2 Use case #1: Root cause identification of Telco cloud
failures
5.2.1 Introduction
As the scale of Telco cloud equipment increases, the difficulty in rapidly locating and addressing faults in Telco clouds
is continuously rising, making it challenging for traditional manual operation and maintenance modes to meet the high-
reliability requirements. In response to this situation, large models can be leveraged to enable intelligent fault location
identification and automated processing for Telco clouds, thereby reducing fault handling time.
5.2.2 Actors and roles
Table 5.2.2-1 describes the use case actors and roles.
Table 5.2.2-1: Use case #1 actors and roles
# Actor and role Description
1 Operator The user who submits the Telco cloud fault diagnosis request.
2 Large model application An application leveraging large models for fault diagnosis fault localization and
resolution and root cause analysis in Telco cloud management.
3 Large model A type of artificial neural network with a complex structure and large-scale
parameters, facilitating intelligent processing and analysis to support fault diagnosis
in Telco cloud management.
4 Operations and A system within NFV-MANO, responsible for raising alarm data queries. It enables
maintenance system the retrieval of alarm data related to fault events in Telco cloud management.
5 Fault diagnosis knowledge A knowledge base storing fault operation and maintenance information (e.g. based
base on manuals) from operators and vendors, including alarm titles, alarm lists, fault
detection methods, and handling recommendations. Used to support fault diagnosis
and root cause analysis within NFV-MANO.
6 Faulty equipment The virtual and physical resources managed by the NFVI that are defective.

NOTE: The knowledge base is owned and managed by the network operator who controls the access.
5.2.3 Trigger
Table 5.2.3-1 describes the use case trigger.
Table 5.2.3-1: Use case #1 trigger
Trigger Description
1 Operator requests to diagnose a Telco cloud fault.

5.2.4 Pre-conditions
Table 5.2.4-1 describes the use case pre-conditions.
Table 5.2.4-1: Use case #1 pre-conditions
# Pre-condition Description
1 Both the large model application and the large model are No additional description.
available.
2 Operations and maintenance system is available No additional description.
3 A pre-built fault diagnosis knowledge base, saving No additional description.
operator and multiple manufacturers' fault operation and
maintenance manuals, mainly including various fault
alarm lists and corresponding fault detection plans and
processing suggestions.
4 Failure in infrastructure causes a fault in the NFVI. No additional description.

ETSI
11 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
5.2.5 Post-conditions
Table 5.2.5-1 describes the use case post-conditions.
Table 5.2.5-1: Use case #1 post-conditions
# Post-condition Description
1 Root cause is identified and fault handling suggestions are The large model application has identified the root
provided. cause of the Telco cloud failure and provided the
operator with the relevant fault handling procedures
followed or notified the operator if the root cause
could not be identified.
5.2.6 Flow description
Table 5.2.6-1 describes the use case flow.
Table 5.2.6-1: Use case #1 flow description
# Actor/Role Action/Description
Begins when Operator -> Large The operator submits a fault diagnosis request in natural language, such as
model application "The alarm count has increased in resource pool A, diagnose and identify the
root cause." The request is sent to the large model application.
1 Large model
The large model application processes the request using the large model,
application -> Large extracting key parameters required for alarm data queries and fault analysis.
model
Large model Using the extracted parameters, the large model application queries the
application -> operations and maintenance system, an NFV-MANO component, to retrieve
Operations and alarm data (e.g. alarms triggered within a specific time range before the fault
maintenance system occurred). This interaction can rely on standardized NFV-MANO alarm object
types (see ETSI GS NFV-IFA 045 [i.6]).
3 Large model Based on the alarm name obtained in step 2 (e.g. 'memory overload'), the
application -> Fault large model application searches the fault diagnosis knowledge base to
diagnosis knowledge retrieve the most likely fault types and the corresponding operating
base procedures for fault location.
4 Large model The large model application sends the collected information to the large
application -> Large model, which includes the fault type (retrieved in step 3) and the faulty device
mode information (obtained in step 2). The large model analyses this data and
selects the most likely fault type, then obtains the corresponding operating
procedures for fault localization. The large model integrates this information
with the fault location method.
5 Large model The large model application accesses the faulty NFVI equipment managed
application -> Faulty by NFV-MANO in accordance with the operating procedures for fault

equipment localization obtained in step 4. It interacts with the faulty equipment to
retrieve detailed diagnostic information.
6 Large model The large model application sends the collected data, including the fault type
application -> Large and equipment information, to the large model for analysis. The large model
model determines whether the fault location can be determined. If the fault is
localized, the process proceeds to step 7. If the root cause cannot be
identified, the large model application reanalyse alternative fault location
methods using alarm information. If all possible methods have been tried, the
large model application output that fault location cannot be determined.
7 Large model Once the fault root cause is located, the large model application reports the
application -> root cause and affected equipment to the operator. Additionally, the large
Operator model application retrieves the most relevant fault handling procedures from
the fault diagnosis knowledge base and provides actionable
recommendations. If all diagnostic attempts fail and the fault root cause
cannot be identified, the large model application notifies the operator of the
failure and may suggest alternative diagnostic paths.

ETSI
12 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
5.3 Use case #2: Query of Telco cloud operational metrics
5.3.1 Introduction
Within Telco cloud management, performance and fault management data are typically stored in databases. Querying
these databases to extract and monitor key operational metrics often involves complex rule definitions. By leveraging
large model technology, query rules can be described in a semantic manner, enabling flexible and context-aware queries
for extracting and summarizing operational monitoring indicators. This approach allows operators and administrators to
retrieve relevant monitoring data and generate summary reports without directly interacting with database query
interfaces or writing query scripts. Consequently, this reduces the technical complexity of Telco cloud operations and
maintenance, while improving efficiency, automation, and intelligence in monitoring and troubleshooting workflows.
5.3.2 Actors and roles
Table 5.3.2-1 describes the use case actors and roles.
Table 5.3.2-1: Use case #2 actors and roles
# Actor and role Description
1 Operator The user who submits the query for Telco cloud operational metrics.
2 Large model application Leveraging large models for processing operator queries related to operational
metrics.
3 Large model A type of artificial neural network with a complex structure and large-scale
parameters, enabling the processing of query data and generation of insights for
operational metrics.
4 Metrics names knowledge A knowledge base storing a list of standard and alternative metric names, along
base with their relationships, used for efficient querying and identification (e.g. user input
like "hard disk" or "storage volume" is mapped to the system's database terminology
such as "hard disk".
5 Operational knowledge A knowledge base within the Telco cloud management domain that stores key
base operational data, such as performance indicators and resource utilization,
supporting fault detection and performance management.

5.3.3 Trigger
Table 5.3.3-1 describes the use case trigger.
Table 5.3.3-1: Use case #2 trigger
Trigger Description
1 Operator requests to query the Telco cloud operational metrics.

5.3.4 Pre-conditions
Table 5.3.4-1 describes the use case pre-conditions.
Table 5.3.4-1: Use case #2 pre-conditions
# Pre-condition Description
1 Both the large model application and the large model are No additional description.
available.
2 Operational knowledge base is available. No additional description.
3 Pre-built metrics names knowledge base, storing synonym lists No additional description.
for various metrics names, primarily converting metrics names
described in users' natural language into system-recognizable
metrics names.
ETSI
13 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
5.3.5 Post-conditions
Table 5.3.5-1 describes the use case post-conditions.
Table 5.3.5-1: Use case #2 post-conditions
# Post-condition Description
1 Providing operational metric query results Providing the corresponding operational metric data
and summary reports according to the operators
request.
5.3.6 Flow description
Table 5.3.6-1 describes the use case flow.
Table 5.3.6-1: Use case #2 flow description
# Actor/Role Action/Description
Begins when Operator -> Large The operator submits a query request in natural language, specifying the
model application type of metrics, resource pools, filtering rules, and time range. The request is
sent to the large model application.
1 Large model The Large model application sends to the large model the metrics type, time
application -> Large range, and filtering rules based on keywords in the operator's natural
model language request, translating it into the relevant parameters for database
query commands.
2 Large model The large model application uses the metric type parsed in step 1 to query
application -> Metrics the metrics names knowledge base, retrieving the corresponding
names knowledge standardized metric names.
base
3 Large model The large model embeds the parsed metric parameters from steps 1 and 2
application -> Large into the specific query parameters of a query command to generate a
model database query command for operational metrics.
4 Large model The large model application uses the generated query command to retrieve
application -> the requested operational data (e.g. CPU, memory usage) from the
Operational operational knowledge base.
knowledge base
5 Large model The large model application passes the query results along with additional
application -> Large prompt words (e.g. report generation requirements) to the large model for
model processing the data and generating a report.
6 Large model The large model application returns the metrics report to the operator, which
application -> includes a natural language or graphical summary of the operational metrics,
Operator along with any insights gained from the process followed.

5.4 Use case #3: Smart deployment plan generation for Telco
cloud
5.4.1 Introduction
The deployment plan details the deployment specifications, storage information, network configurations, and other
relevant components. In the NFV context, this includes the configuration of VNFs, their interconnections, and the
provisioning of NFVI resources, such as compute, storage, and networking. The plan also outlines the configuration of
both VNFs and NFVI resources, to meet high-availability and performance requirements. In addition to initial
deployment, the plan also covers configuration changes needed throughout the lifecycle, ensuring that the system adapts
to evolving network requirements.
ETSI
14 ETSI GR NFV-EVE 027 V6.1.1 (2026-04)
However, the strict high-availability requirements of the deployment plan also present a significant challenge and a high
technical barrier for operators to fill out. In response to this situation, large model technology, which supports
interaction with users in natural language, extracts configuration parameters required for cloud deployment through
multiple rounds
...