Automating network slicing

operations in the 5G core

with orchestration
Maximizing speed, efficiency and quality in
lifecycle management using dynamic automation

White paper
 Contents

Introduction 3
Crucial challenges 3
Technology focus areas 4
The value of network automation 4
Mastering network complexity 4
Rapid service delivery 5
Addressing operational efficiency with automation and orchestration 6
Orchestration and intent-based automation 8
Dynamic automation: smart templating and code auto-generation 9
Using CI/CD and a DevOps approach for more agile packet core operations 10
To slice or not to slice: That is not really the question 11
Management of core network functions 11
Network mediation 11
Network and slice assurance 12
Nokia Network Services Platform 13
Conclusion 14
Appendix: 3GPP architectures for the core and the orchestration hierarchy 15
5G network slicing in 3GPP 15
Network automation and orchestration hierarchy in 3GPP 16
Abbreviations 17

2 White paper
Automating network slicing operations in the 5G core with orchestration
Introduction
The number of 5G connections is projected to reach 1.8 billion by 20251. Current forecasts indicate that
by this point, 5G will account for approximately 30 percent of connections in markets such as China and
Europe, and about half of all connections in the US. Overall mobile data traffic will grow at an estimated
annual rate of around 55 percent. The impacts of this growth on the mobile industry and its customers
(consumers, institutions and industry) will be profound.
While the continued optimization and proliferation of consumer broadband services such as enhanced
mobile broadband (eMBB) and fixed wireless access (FWA) is vitally important, much of the impact of 5G
will be in the enablement of innovative Industry 4.0 and public sector use cases. New opportunities will
be addressed, and existing businesses and institutions will be digitalized, automated and transformed
as we enter the next technological and industrial revolution2. Massively scaled Internet of Things (IoT)
applications will coexist with use cases in areas such as public security, autonomous transportation,
telemedicine, distance learning, rapid-response logistics, high-efficiency manufacturing and optimized
energy distribution.
It is not an exaggeration to say that these trends represent important transitions in the way society at
large will function. They also present an opportunity for communications service providers (CSPs) to bring
even more value to the connected society. With the rise of distributed, cloud-native networks, this value
resides in the network and provides the platform for web applications.
The 5G network extends over a set of domains from the radio access network (RAN) to the core. This
paper focuses on the 5G core and the challenges and solution approaches relating to the crucial topics
of automation and network slicing.

Crucial challenges
Agility
Rapid instantiation of existing, and innovation of new, services is required to address disparate use cases.
Cloud service providers have set expectations for fast, “one-click” turn-up times. A range of use cases
(many currently unforeseen) will be supported on a single infrastructure. The network must be able to
flexibly and equitably apportion resources to support service-level agreements (SLAs) that are appropriate
to each of the addressed use cases.
Elastic scalability
The network must be able to scale up and out to accommodate traffic surges and adapt to continuously
evolving customer demand and real time events and emergencies. Scaling must be fast and, where
possible, anticipatory to prevent service interruptions or degradations.
Cost control
High operational efficiency is needed across the entire lifecycle of platforms and services. Network
resources such as compute and connectivity must be carefully engineered to build the best infrastructure
for the least cost.

1 GSMA, “The Mobile Economy 2020,” London, United Kingdom, 2020.

2 M.K. Weldon, The Future X Network: A Bell Labs Perspective, CRC Press, 2016.

3 White paper
Automating network slicing operations in the 5G core with orchestration
Technology focus areas
The 5G core architecture
The 5G core architecture features several innovations that address the challenges above. For example,
network slicing provides the ability to create a set of independent, end-to-end logical networks that run
on the shared physical infrastructure (including the core and other domains). The network slices must be
rapidly deployable and manageable with automation. They must provide strict, tailored service quality to
each client group. Control/User Plane Separation (CUPS) permits the optimal placement of control and
real-time transport processing to meet performance targets and allow independent, efficient engineering
of resources to support these functions.
Cloud-native philosophy
The 5G core will be installed on modern cloud platforms in private and public clouds, rather than
server-based facilities with heavy administrative requirements. Cloud-native principles include modular
decomposition of functions into microservices to partition enhancements and analysis, containerization
of functions for portability and geo-distribution, and continuous integration/continuous delivery (CI/CD)
for reliable and rapid development and deployment. Containerization provides accelerated interoperability
within the broader operational software stack. It also supports a DevOps culture to enable rapid delivery
of feature upgrades and validation. In addition, cloud-native design builds in the required resiliency and
scalability for supporting independent network functions as microservices, and the flexibility to leverage
third-party and open-source software.
Automated network operations
Automated operations that extend throughout the lifecycle of platforms and services are vital for efficient,
reliable, secure and successful network management. Automation includes concepts such as abstraction,
where only the necessary attributes of a layer or component are exposed on a programmable interface.
Abstraction simplifies operations and partitions regions of expertise to allow for intent-based automation,
selectively shielding system complexity from operators and/or higher-level systems. Operational expertise
and design details can be captured and coded into templates, workflows and Network Service Descriptors
(NSDs) to speed and standardize repetitive tasks and best practices.

The value of network automation

Automation and an associated automation strategy support the deployment of the 5G core components
in a cloud-native environment and effectively unlock their full potential. These are key 5G investments
that have the potential to provide significant returns in network efficiency, competitiveness and financial
performance. Automation is indispensable for operating successfully in a competitive business context.

Mastering network complexity

Automation eases the management of increasingly dynamic and complex networks throughout the
lifecycle of network elements and functions and network slices.
A Nokia Bell Labs study3 highlights the influence of automation on the transformation of the network
and its effect on total cost of ownership (TCO), as illustrated in Figure 1.

3 N. Raman et al., “Future X Network cost economics: A network operator’s TCO journey through virtualization, automation, and network slicing,” Bell Labs Consulting,
2018.

4 White paper
Automating network slicing operations in the 5G core with orchestration
Figure 1. Efficiently addressing scale and speed using automation

Manually sliced
network
TCO TCO +30%
Virtualization
Parallel platforms
Manual slicing
Initial virtualized
network
TCO +19%
Virtualization
Parallel platforms Automated sliced
Classical network
physical TCO –32%
CSP network Virtualization
No parallel platforms
Automated slicing

Degree of Degree of
transformation automation

Nokia Bell Labs estimates that the added complexity of performing tasks manually (e.g. slicing) in
virtualized networks will increase TCO by 30 percent compared to traditional networks. The key takeaway
from this study is that automation is the only way to make network slicing economical while simultaneously
providing rapid service delivery.
The curve in Figure 1 shows how TCO changes with maturity steps as the entire network evolves.
Introducing the right automation strategy earlier will make the curve shorter and shallower and increase
business competitiveness.

Rapid service delivery

Automation accelerates service delivery and enables rapid turn up of new or modified network slices. The
automated implementation of network slicing can deliver highly customized networks tailored to individual
customers’ specific requirements in a cost-effective, timely and efficient way. Operators are looking to
deploy network slicing solutions now to tap into the business-to-business opportunity offered by 5G, for
example, in manufacturing, enterprise campuses, smart cities, transportation, logistics and healthcare.
The slices are each governed by an individualized SLA. Figure 2 shows a typical deployment of network
slices across an end-to-end 5G network.
A key success factor will be the ability to quickly meet customer needs (agility and quality drivers) in an
economical way (cost driver). However, the complexity involved in managing many slices and customers
threatens the business model as demonstrated in Figure 1. Implementing extreme automation streamlines
operations, brings down the operational costs of network slices and allows service velocity expectations
to be met.

5 White paper
Automating network slicing operations in the 5G core with orchestration
Figure 2. End-to-end network slicing in a 5G network

E2E network slice orchestrator

E2E network slices
RAN slices Transport slices Core slices
Automobile
Tenants manufacture (AM1) RAN slice Transport Core slice
Infotainment controller slice controller controller(s)
AM1 AM2
HD maps infotainment HD maps Public
Automated driving and
private
Automobile clouds
manufacture (AM2)
Automated driving

Public safety (PS) RAN Core

Operator-X
Video surveillance
Multi-source PS video
artificial surveillance
E2E network slices: NS1 NS2 NS3 NS4 NS5 intelligence

Addressing operational efficiency with automation and

orchestration
Orchestration is the supervisory process of automatically programming the network across a set of
domains to support application and service specifications. It involves lifecycle management tasks such
as creating, deploying, updating, terminating and deleting physical, virtual and containerized network
functions, along with all the networks that connect these network functions.
The automation made possible by the Nokia 5G cloud packet core (CPC) domain orchestrator was recently
analyzed by a Tier 1 CSP and found to deliver a 71 percent return on investment (ROI) over five years
with a payback period of three years. The comparison for this analysis was an approach that used next-
generation OSSs.

Tier 1 communication service provider saw a 71% ROI over 5 years with a payback period of 3 years.

Figure 3 shows the hierarchy of automation and orchestration entities engaged in end-to-end network
slicing. The 3GPP standardization of this hierarchy is described in more detail in the appendix.

6 White paper
Automating network slicing operations in the 5G core with orchestration
Figure 3. Automation hierarchy of end-to-end network slicing

Ecosystem

Customer portal Customer

request

E2E slice orchestration

Operations

Lifecycle
RAN domain Transport domain Core domain
management of
slices and network
functions

Network

Multi-technology Multi-domain Multi-vendor

The establishment of a slice is simplified by the structure and information abstraction of the automation
hierarchy:
• The customer or operator interacts with a relatively simple and straightforward portal without any
knowledge of the complex networks that must be programmed to complete the action. All that needs
to be passed at this interface is some information about when and where the slice is needed, along with
basic network-related parameters such as latency, bandwidth and desired reliability.
• The end-to-end slice orchestration function is aware of the domain layout and the available end-to-end
service types and passes the appropriately augmented requests to the domain orchestrators. The 3GPP
calls this the network slice management function (NSMF).
• The domain orchestrators contain deep expertise in the technology that they control. Their role is
to implement the slice in their domain by orchestrating the detailed programming of objects, such
as network functions and interconnecting subnets. The 3GPP calls this the network slice subnet
management function (NSSMF). Management tasks relating to network functions within the slice are
carried out by the network function management function (NFMF).
The architecture shown in Figure 3 embodies several attributes and characteristics that enable efficient
operations across the complete lifecycle of network slices, including:
• Split of responsibility: Network slicing is a complex, multi-dimensional problem that requires a
separation of responsibilities. The slice orchestrator requires a simplified view of the network domain
service so that it can focus on the business-level intent. Each domain requires network expertise
(technology, protocols, architecture) to accomplish the detailed implementation. This is achieved
through intent-based network operations.

7 White paper
Automating network slicing operations in the 5G core with orchestration
• Dynamic automation: Automation should be promoted as much as possible to reduce coding effort
for network designers and accelerate operations over the complete lifecycle, including the creation,
updating and deletion of network slices. Auto-generation of workflows using deployment templates
and appropriate data modeling languages takes basic automation to a more dynamic paradigm that
can eliminate many manual tasks such as script writing.
• Modularity of architecture: A modular or plug-in-based approach is highly desirable throughout the
architecture. It allows for rapid, continuous integration and deployment of new or updated network
functions and other components as needed without a major re-release of the whole stack or any
undesirable impacts on other components or layers.
These techniques and practices foster a DevOps culture. This allows changes in domain function or in
slice attributes to be rapidly moved to the production environment with minimal and bounded risk.

Orchestration and intent-based automation

Intent-based automation shields against complexity by supporting declarative programmability that
separates the details of the “how” (needed for implementation) from the “what” (business-level intent).
At the end-to-end slice orchestration level, the focus is on business intent rather than the network,
network function and associated details. This abstraction hides unnecessary domain-specific complexity
from the consumer of a slice, thereby simplifying slice consumption.
Referring to Figure 3, intent-based automation should be implemented through the entire software
stack – from business and operational systems to end-to-end service orchestration and domain network
orchestration – so that intents can be programmed for each software layer and intent-chaining can be
used. In this fashion, the integration of systems through standardized APIs can drive the necessary
commands and transactions as designed and implemented with appropriate subject-matter expertise
at each layer.
This intent-based design and implementation approach enables multiple groups and departments
within an organization (each with different concerns and systems knowledge) to work together more
harmoniously and synergistically, while allowing appropriate focus on the responsibilities of each group.
The results for the business are more agile service operations and reduced OPEX.
One example of a resourcing benefit is that IT/OSS workflows can be defined by IT/OSS personnel who do
not require domain-specific knowledge of the underlay network and associated technologies. This subject-
matter expertise will be provided by the network engineering teams that own the 5G network domains. It
will be appropriately deployed in models and workflows associated with the domain controllers.
For smaller operations with limited staffing, adaptions to end-to-end and cross-software-stack workflows
will become easier and more scalable to develop with an approach that ensures consistent delivery and
error avoidance for every team member. For example, intent-based templates can support configuration
auditing by ensuring that intents are being achieved. They can also help ensure consistency by notifying
operators or automatically triggering closed-loop changes when deviations occur.
For more information on intent-based automation please read our application note on “Network
automation and programmability: Using the Nokia NSP for intent-based networking (IBN).”

8 White paper
Automating network slicing operations in the 5G core with orchestration
Dynamic automation: smart templating and code auto-generation
An initial core slice deployment is a lengthy and complex process with numerous steps. These include
creating the cloud infrastructure, selecting, instantiating and configuring the network functions, and
establishing data center connectivity for all components. These operations seem relatively amenable
to automated workflows. As in most software systems, the complexity arises in exception handling. For
example, script errors and version upgrades to various components can all complicate a pure scripting
approach and cause long delays.
There are opportunities to make automation more dynamic and hence more efficient. As a best practice,
it should be possible to greatly reduce coding efforts by auto-generating the workflows, which can quickly
resolve the barriers to completion. This auto-generation should extend to interworking with different
domain systems, including cloud and network management systems and controllers, and make use of
deployment templates, called NSDs.
NSDs provide standardized methods for describing network services, their components, their relationships,
and the processes that manage them. These templates define and drive the lifecycle management of
the constituents of the slice (network functions, configurations, networks and infrastructure) in the core
domain. The value of NSDs is to minimize manual coding efforts by capturing this design information
and expertise. Upon execution, the templates automatically generate the associated scripts, workflows,
artifacts and API calls, which may represent many thousands of lines of code. This activity carries out the
desired lifecycle function(s).
Complete lifecycle management is an important attribute of dynamic automation. In addition to the
initial instantiation and configuration of a core slice, updates to change the slice characteristics should
be highly and dynamically automated. For example, it should be possible to analyze these slice updates
automatically to determine whether an existing network function needs to be reconfigured or a new one
needs to be created. If the existing set of templates that describe the slice are comprehensive enough,
the new update should be able to be translated automatically into the required workflows to delete, create
and configure as required. Without this capability, a massive script-writing exercise would be needed to
accomplish updates and deletions. This process would be time consuming and error prone.
NSDs often use Topology and Orchestration Specification for Cloud Applications (TOSCA), a declarative
language based on YAML, a human-readable markup language. However, the different standards and
approaches used for modeling specifications and programming languages across domains create design
and implementation challenges.
For example, most network functions do not support TOSCA but do support YANG. As an IETF standard,
YANG has become a prevalent requirement as a network modeling specification for configuration of the
network because network engineering specialists with expert knowledge of the network functions prefer
it. In contrast, data center network architects prefer to use OASIS-standardized TOSCA to define cloud
network orchestration, whereas YANG is not well known within this community.
Some within the industry have attempted to align the data center installation intents and the configuration
of the network functions to create a uniform YANG-based modeling approach. These attempts have failed,
essentially because the underlying paradigms for YANG and TOSCA are different. YANG assumes a strict
hierarchical structure of the data models, whereas TOSCA, which is based on IT service-based architecture
(SBA) principles, allows for any-to-any and dynamically changing relationships between objects. Trying to
force any possible object relationship into strict YANG hierarchies has proven to be a very labor-intensive
and therefore expensive and time-consuming task because automation is not possible. Therefore,
the recommended approach is to use both TOSCA and YANG as complementary intents to realize the
envisaged automation.

9 White paper
Automating network slicing operations in the 5G core with orchestration
In summary:
• Dynamic automation – that is, the automatic generation of critical workflow code – will make it
significantly easier and faster to incorporate network domain changes into cloud network orchestration
workflows. Capturing network service designs into intent templates, and then managing the lifecycle
of these templates allows for the automated generation of the code. This significantly reduces the
programming effort in creating, deleting and efficiently updating cloud network and network function
resources.
• For optimal instantiation and configuration of network functions, core network orchestration and
management software must be able to support both YANG and TOSCA. This allows the software to
efficiently carry out a complete range of operational activities throughout the lifecycle of an end-to-end
network slice. If the software lacks the flexibility to support both languages, it can lead to considerable
work in repeated manual translations.

Using CI/CD and a DevOps approach for more agile packet core operations
Mobile operators aspire to automate core network operations throughout the lifecycle of services and
software to become more agile, responsive and efficient. Until recently, however, operators have been
slow to reach this level of automation. Many are still using manually intensive and siloed approaches across
teams. As an example, software update mechanisms must guarantee that any changes can be introduced
into the network in a way that is agile and efficient and results in minimal or no disruption to the operation
of existing core network functions.
The optimal approach to accelerate the onboarding of software upgrades is to use a consistent, complete
automation strategy according to lifecycle management principles and best practices for CI/CD. As a result,
a new software upgrade, deletion or update for a specific network function can be integrated and deployed
efficiently, smoothly and continuously into a production environment that contains up to several thousand
packet core network functions. This is greatly facilitated by a containerized architecture where applications
such as network functions are cloud-native.
CI/CD allows operators to rapidly introduce enhancements to network functions, test them in controlled
environments and put them into service. This can only be achieved by identifying, confronting and
resolving operational complexity using templating, complemented by scripts and workflows. The same
template-driven philosophy that automates slice lifecycle management can be applied to tasks such as
the onboarding of software upgrades. To support effective CI/CD, all exception conditions (or error paths)
in workflows must be anticipated and automated mitigation strategies must be in place – for example, to
automatically undo and roll back an update if blocking difficulties are encountered.
Investments in CI/CD will provide improvements in agility, allowing operators to continuously enhance
network functions and efficiently deliver new features to the production environment. CI/CD is part of a
broader DevOps culture that will facilitate the rapid development, deployment and operation of new 5G
service offerings.
An effective DevOps culture seeks to boost the efficiency and reactivity of an organization. Automated
testing strategies with transparency between the operator and vendor(s) allow optimal coverage of
scenarios to simulate the end-to-end production network. This testing is often carried out in a testbed or
sandbox environment. Value is seen in transitioning to a ‘canary testing’ model where a carefully backed-up
production site is used for test cycles with live traffic.

10 White paper
Automating network slicing operations in the 5G core with orchestration
To slice or not to slice: That is not really the question
5G network slicing in the 5G core is a primary driver for the automation of operations. However, while
slicing as a vision is uncontested in the industry, the exact timing of large-scale slicing deployment is
not clear. On the other hand, the significant trend towards CI/CD and DevOps also drives the need for
automation and is doing so with urgency. CI/CD and slicing are both technologies that meet important
business needs and have a crucial element in common – automated systems, which are necessary to
facilitate the operations of, and service delivery over, networks. Network slicing, in particular, is unlikely
to be deployed in volume until fully automated operations are available.
Nokia has therefore architected its products with a vision of slicing that has CI/CD and operations
automation as founding principles. This approach is intended to remove the need to write and re-write
operations automation scripts if they can be automatically generated. Correspondingly, CI/CD and
operations automation support systems and artifacts are built with slicing in mind. The goal is to establish
CI/CD capability on a consistent capability roadmap to support the lifecycle management of slicing and be
ready to scale efficiently as the volume of slicing deployments starts to ramp up.

Management of core network functions

CPC network function management continues to be a crucial factor in achieving efficient network
automation. The NFMF plays a critical role in supporting automation functions with the NSSMF in the
context of specific 5G CPC subnet slices. Key capabilities include network mediation and assurance.

Network mediation
The NFMF should provide communications with a multivendor framework for model-driven mediation
(MDM) to support the rapid deployment of new network function software releases and service models.
With MDM, NFMF upgrades to support new network functions can be decoupled from full CPC NFMF
upgrades. It can also inherently provide forward compatibility for supporting new network functions and
related service models without requiring a platform upgrade.
This new paradigm shift to MDM is fundamentally different from the present mode of operations and
delivers dramatic improvements. In the past, for example, operators may have had to wait many months
for new network function releases to be supported in an element or network management system (EMS/
NMS). This occurred because vendors or in-house designers needed to change network function feature
support and the necessary network function and service object models in the code base of these systems.
Operators often simply had to wait for new versions of vendor systems to get support for new network
function releases. In addition, the deployment timeline for vendor software needed to be planned,
implemented and tested for platform-wide upgrades. This could add many more months of delay,
especially if OSS integrations also needed to be re-validated.
With MDM, an NFMF can significantly reduce these deployment delays – to days or hours in many cases,
depending on the project scope. It can efficiently expose new device features to northbound systems
by adopting new southbound and northbound models and creating new adaptation scripts to translate
between the two. There is no longer a need to change internal models.
An NFMF that uses MDM can enable the maximum level of automation through YANG modeling, which has
become predominant in modern CPC networks. YANG models can be hot-deployed, so NFMF support can
be ready to use as soon as the YANG model is made available and deployed using MDM. This is possible
because the object models and NFMF support are automatically derived from the YANG model.

11 White paper
Automating network slicing operations in the 5G core with orchestration
Out-of-the-box support should also be provided for many standardized YANG models, such as the IETF
L2 and L3 service models. In addition, GUIs such as those for model-driven configuration and RESTCONF
northbound APIs can be autogenerated from these YANG models. MDM and model-driven automation
are key underlying features that enable intent-based management, an advanced best practice for
implementing an NFMF for the CPC. A NFMF’s model-driven API can also be further enhanced with
programmable workflow management to support fine-grained network and service control.

Network and slice assurance

To make CPC automation work well in live network deployments, an NFMF must support the full operations
lifecycle, including assurance that provides real-time monitoring, visualization, reporting and analytics on
5G CPC slices. Network assurance data must be available for each network function and network slice to
enable advanced fault and performance management capabilities that can identify, prevent and mitigate
issues before they impact 5G services.
Network assurance must also enable programmability for implementing closed-loop assurance policies to
automate remediation and optimization. These include monitoring policies that trigger actions to scale
out, reconfigure, move network functions or find alternative paths to connect a network function when
specific thresholds are crossed. If the NFMF cannot find a solution to prevent an SLA violation, it should
report the problem to the NSSMF.
The NFMF will receive telemetry data continuously from the CPC network for all resources and aggregate
them into key SLA adherence metrics. These aggregated key metrics form part of the real-time CPC
slice SLA, which will be used for various analytics inside the core NSSMF, as well as for enabling various
types of reporting on transport slices. The key metrics will also be sent to the end-to-end network slice
orchestrator for further analytics and aggregation. To reduce mean time to repair (MTTR) and maintain
SLAs, the NFMF should augment CPC assurance and monitoring to provide dynamic supervision and
reporting capabilities as a best practice. As shown in Figure 4, these capabilities should help network
operations staff understand:
• What is happening in the network in the present, in near-real time
• What has happened in the network in the past
• What will likely happen in the network in the future, based on predictive analysis

Figure 4. Augmenting slice assurance and monitoring to understand past, present and future

Past Now Future Time

What happened? What’s happening? What will happen?

Reporting Supervision Predictive analytics

Understand Root cause analysis

Fix Troubleshooting Optimization Prevent

Benefits
Slice • Reduce MTTR
• Maintain SLAs
• Accelerate problem resolution

12 White paper
Automating network slicing operations in the 5G core with orchestration
An NFMF should provide efficient tools that enable operators to quickly troubleshoot issues with the CPC,
provide root cause analysis, and bring visibility to current and past (fixed) issues. It should also enable
operators to make changes to prevent incidents before they occur and impact SLAs.
A best practice for optimizing assurance and monitoring for 5G slices is to use NSSMF and NFMF data to
implement CPC slice operations applications. For example, CPC health dashboards can provide a summary
of the overall health of all CPC slices in the network and allow operators to drill down into health views of
individual slices.
Another important assurance capability for CPC health dashboards is to allow operations to seamlessly
cross-launch adjacent assurance applications for problem investigation and troubleshooting. For example,
5G CPC dashboards should allow for cross-launching of other assurance applications, such as for network
supervision, service supervision, fault management, analytics and reporting.

Nokia Network Services Platform

The Nokia Network Services Platform (NSP) is a 5G core network orchestrator (3GPP NSSMF) and CPC
network function management (3GPP NFMF) operations platform. It is also a transport domain controller
and network manager for IP, MPLS, optical, PON and microwave networks.
NSP orchestrates network slicing across 5G transport and core network domains. It automates the entire
lifecycle management of network slices (as well as the individual network functions in each slice) across
these two domains to simplify network operations and network design tasks:
• For operations teams, NSP offers complete, automated lifecycle management of slices in each domain.
This includes the creation, deployment, configuration, connection, monitoring, update and deletion of
networks and network functions in a slice. NSP also optimizes and assures slices in the transport and
core networks to minimize resource utilization while meeting strict performance requirements under
continuously changing conditions.
• For network engineering and architecture teams, NSP provides built-in domain-level network design
expertise across transport and core networks and offers the ability to eliminate coding efforts by auto-
generating comprehensive programmatic logic. These auto-generation capabilities leverage deployment
templates with appropriate intent-based modeling languages (YANG and TOSCA).
In addition, NSP offers both teams full visibility of slicing across the transport and core domains. This
cross-domain visibility will enable these teams to make smarter decisions about network architecture,
deployment and configuration.
Nokia subject-matter experts hold leadership positions within IETF, ETSI and 3GPP. In these roles, they
help drive high-quality industry standardization and ensure strong compliance across the NSP.

13 White paper
Automating network slicing operations in the 5G core with orchestration
Conclusion
Automating the operation of the 5G core domain with network orchestration maximizes speed and
efficiency in the design, deployment, assurance and management of 5G core networks and slices. It
enables operators to gain:
• Agility: Operators can quickly deploy and update core slices that answer end customer expectations in
terms of service velocity.
• Quality: Operators can offer core slices that enable their customers to run services with ultra-reliability
and low-latency requirements.
• Cost savings: Automation can reduce the operational cost of slicing by 30 percent4. It enables operators
to optimize the usage of their (shared) network infrastructure while committing to strict SLAs.
• Simplicity: Operators can master the complexity that virtualization and slicing bring with more
abstraction and automation.
Nokia recommends a complete dynamic automation strategy for the 5G core, including auto-generation
of workflows and scripts using templates, automated onboarding of software upgrades and explicit
application of lifecycle management principles. The use of templates ensures efficiency and repeatability
of desired outcomes. It should leverage the TOSCA and YANG modeling languages as appropriate,
complemented with other open-source tools as needed. This strategy supports full automation across
the end-to-end lifecycle in a DevOps culture, implementing the principles of CI/CD.
In the core, Nokia recommends automation capabilities that enable comprehensive assurance of 5G
packet core network functions. These capabilities include visualization of slice and network function
health based on data collected, OAM tests and events such as faults and threshold crossings. They
also include advanced analytics and correlation to provide insights that accelerate root cause analysis.
These automation capabilities can also trigger closed-loop optimization to reconfigure or move network
functions, or find new paths to connect them, if problems or SLA violations occur.
Nokia supports its 5G core automation strategy with proven domain expertise. To date, more than 100
customers have deployed the Nokia packet core. Nokia also supports massive deployments in transport
networks and RANs. This end-to-end experience is invaluable in translating strategic technology plans into
successful business outcomes with comprehensive products backed by an extensive global professional
services capability.

4 N. Raman et al., “Future X Network cost economics: A network operator’s TCO journey through virtualization, automation, and network slicing,”, Bell Labs Consulting,
2018.

14 White paper
Automating network slicing operations in the 5G core with orchestration
Appendix: 3GPP architectures for the core and the
orchestration hierarchy
5G networks are being architected to address the need for more dynamic cloud network services. This is
intended to accelerate both service innovation and deployment velocity. Evolving to the 3GPP 5G SBA, as
shown in Figure 5, supports progress towards these goals.

Figure 5. CPC evolution to 5G core network functions

Introducing standard API

and HTTP2.0/TCP over a UDSF UDR CHF PCF SCP BSF SEPP AUSF
common control bus

Service-based architecture

UDM SMSF NEF NSSF NRF N3IWF AMF SMF UPF

Existing core architecture
with many rigid and inflexible
reference points
CPC-provided 5G core network functions

Evolution to 3GPP 5G SBA standards facilitates advanced capabilities such as network slicing and CUPS to
support the more efficient delivery of a rich suite of subscriber services.
More coordination and orchestration for automating the core network, along with establishing transport
network connectivity that will meet 5G SLA requirements, will be essential for delivering on the promise
of new 5G services, especially for ultra-reliable low latency communication (URLLC). In particular for
5G network slicing, the core domain network orchestrator, also known as the 3GPP NSSMF, is a critical
component for automating and orchestrating 5G core network functions. For example, the CPC-provided
5G core network functions, shown in blue within Figure 5, will be the initial priority for supporting 5G slicing
orchestrated by an NSSMF. Other 5G core network functions, shown in gray within Figure 5, may be more
static for the foreseeable future. In most cases, they are not required to be instantiated on the fly in the
initial phases of slice-based offerings using an NSSMF.
The end-to-end network slicing orchestrator, also known as the 3GPP NSMF, interworks with the NSSMF
for the 5G core, and corresponding NSSMFs for other domains, to enable holistic 5G service orchestration.
This orchestration is critical for achieving efficient end-to-end 5G operations. The following section
describes the hierarchy of these management functions.

5G network slicing in 3GPP

The standardization of interfaces with orchestration for data models and APIs plays an important role in
enabling consistent automation practices. 3GPP R16, released in the third quarter of 2020, is the leading
source of specifications for 5G network slicing for the communications industry. The formal 3GPP version
of the automation hierarchy introduced in Figure 3 is shown in Figure 6.

15 White paper
Automating network slicing operations in the 5G core with orchestration
The communication service management function (CSMF) acts as a gateway for translating customers’
communication service-related requirements into network slice-related requirements for the network
slicing ecosystem. Figure 6 shows that with the 3GPP 5G architecture, the CSMF interacts with an NSMF.
This NSMF interacts with domain specific NSSMFs for the RAN, transport and core, respectively.

Figure 6. 3GPP 5G management/orchestration hierarchy and architecture

Communication Service Management Function

(CSMF)

Network Slice Management Function

(NSMF)

3GPP API 3GPP API

RAN Network Slice Subnet Transport Network Slice Subnet Core Network Slice Subnet
Management Function Management Function Management Function
(RAN NSSMF) (Transport NSSMF) (Core NSSMF)

RAN Network Function Transport Network Function Core Network Function

Management Function Management Function Management Function
(RAN NFMF) (Transport NFMF) (Core NFMF)

3GPP APIs should include support for requirements outlined in technical specifications such as 3GPP TS
28.530, 531 and 532, for management and orchestration use-cases and requirements, as well as 28.540
and 28.541, for the definition and fulfillment of requirements towards a network resource model (NRM)
for network resources, next-generation RAN, 5G core network and network slices.
Domain-specific NSSMFs also interact with their domain-specific NFMF counterparts to provide complete
network automation and full lifecycle management. For example, during the 5G slice instantiation
automation process, the design and creation of the end-to-end network slice needs to divide the end-
to-end slice SLA into slice subnet SLAs and network resource requirements for each domain. Each NSSMF
will perform the network slice subnet instantiation/activation and SLA application for its own domain.
In the core domain this means that the NSSMF interacts with an NFV orchestrator (NFVO) or container
orchestrator and cloud stack software to allocate and instantiate network function resources in virtualized
or cloud-native (containerized) environments as desired.
Through this process, the network functions are instantiated with basic configuration and the slice
subnet SLAs are translated into network function configuration parameters and provisioned by the NFMF.
The NFMF then activates these network functions and the network service associated with the slice by
configuring a service that meets the slice subnet SLA requirements.

Network automation and orchestration hierarchy in 3GPP

Network automation for the CPC network functions in the 5G core should encompass network function
orchestration, management and assurance. A core NFMF that is aware of core subnet slices also improves
automation by enabling operations and management on a per-slice basis and by simplifying integration
with an NSSMF. For example, by programming NSDs to implement an SLA policy on core slice creation,
the NSSMF can interwork with the NFMF to define the optimal engineering of CPC network functions. In
addition, the NSSMF ensures consistency of operations on a per-slice basis for automating core slices
across the core network and into public and private clouds.

16 White paper
Automating network slicing operations in the 5G core with orchestration
Abbreviations
API application programming interface
BSS business support system
CI/CD continuous integration/continuous delivery
CPC cloud packet core
CSMF communication service management function
CSP communications service provider
CUPS control/user plane separation
DSP digital service provider
eMBB enhance mobile broadband
EMS element management system
FWA fixed wireless access
IoT Internet of Things
LCM lifecycle management
MDM model-driven mediation
mMTC massive machine type communications
MTTR mean time to repair
NF network function
NFMF network function management function
NFVO NFV orchestrator
NMS network management system
NRM network resource management
NSD network service descriptor
NSMF network slice management function
NSP Network Services Platform
NSSMF network slice subnet management function
OSS operations support system
RAN radio access network
ROI return on investment
SBA service-based architecture
SLA service-level agreement
SMF session management function
TCO total cost of ownership

17 White paper
Automating network slicing operations in the 5G core with orchestration
TOSCA topology and orchestration specification for cloud applications
UPF user plane function
URLLC ultra-reliable low latency communication
VNF virtual network function
WAN wide area network
YAML yet another markup language
YANG yet another next generation

About Nokia
We create the technology to connect the world. Only Nokia offers a comprehensive portfolio of network equipment, software, services and licensing opportunities across
the globe. With our commitment to innovation, driven by the award-winning Nokia Bell Labs, we are a leader in the development and deployment of 5G networks.

Our communications service provider customers support more than 6.4 billion subscriptions with our radio networks, and our enterprise customers have deployed over
1,300 industrial networks worldwide. Adhering to the highest ethical standards, we transform how people live, work and communicate. For our latest updates, please visit
us online www.nokia.com and follow us on Twitter @nokia.

Nokia operates a policy of ongoing development and has made all reasonable efforts to ensure that the content of this document is adequate and free of material errors
and omissions. Nokia assumes no responsibility for any inaccuracies in this document and reserves the right to change, modify, transfer, or otherwise revise this publication
without notice.

Nokia is a registered trademark of Nokia Corporation. Other product and company names mentioned herein may be trademarks or trade names of their respective owners.

© 2020 Nokia

Nokia Oyj
Karaportti 3
FI-02610 Espoo, Finland
Tel. +358 (0) 10 44 88 000

CID207854 (November)