Building a cloud

native infrastructure
AEricsson’s
guide todual-mode
deploying cloud
5G Core solution executive summary
native platforms for 5G

October 2020
2 Ericsson  |  Building a cloud native infrastructure

Introduction

5G Standalone (SA) drives a rapid introduction of

cloud native infrastructure.

The industry is evolving to the new Figure 1: Ericsson cloud native design principles
standardized 5G Core based on the cloud
native service-based architecture (SBA) Cloud native technology Ericsson cloud native design principles
and New Radio (NR). This new 5G SA
network is needed to address multiple
vertical services for industries and Infrastructure
Microservices DevOps
enterprises, and to open up new business agnostic
opportunities. The areas in focus are
Orchestration Decomposed
manufacturing and automotive, and automation software
private and mission-critical networks, Cloud native
and wireline access. 
Our industry – and the entire world
– has a big opportunity, enabled by low State
Application
latencies, high bandwidths, network CI/CD Containers optimized
residency
design
slicing, and programmable networks.
We also face two big growth challenges:
how to be more efficient and lower the
operational and capital expenditure
costs when data traffic keeps growing, many vendors’ open source software. don’t fly, they can be phased out just
and how to deal with the increasing 5G SBA is a paradigm shift for the core as easily. Simplified operations will
network complexity in 5G. The complexity network architecture of mobile networks. also play a critical role in enabling new
growth is in two dimensions: firstly, the The new 5G Core architecture will advanced 5G services as they will require
number of layers and functions in the make mobile networks more efficient additional network capabilities that will
stacks of hardware, virtualization layers, and easier to operate. It reduces time significantly increase the complexity in
applications and operations support to market for new services and makes it the network. The 5G Core architecture
system and business support system easier to manage services with shorter and the cloud native/microservices
(OSS/BSS). The second is the number lifespan. It will, for instance, allow new implementations will reduce this
of potential combinations within and services to be launched quickly to test complexity while keeping operating
between the layers with products from the market demand and, if the services expenses low.

Figure 2: Drivers for cloud native infrastructure

5G applications CNCF community is New 5G use cases Optimized infrastructure

designed using driving cloud native need distributed for cloud native
cloud native principles across industries cloud infrastucture telco applications
3 Ericsson  |  Building a cloud native infrastructure

The evolution to cloud native

builds on the network functions
virtualization (NFV) transformation

Since 2012, and the start of the virtualization of network

functions, the industry has worked hard to get working solutions,
and we now see live commercial systems that work well.

Some of the networks have a few System-verified NFVI stack limit systems integration, and lower
virtualized network functions (VNFs) on has been a key success factor the cost of maintaining exclusive
smaller or node-based NFV infrastructure The stability of telco cloud platforms stack combinations.
(NFVI) platforms. Others have built larger has been a challenge for the industry The benefits are mainly on the opex side:
horizontal telco clouds, able to run multiple in past years but has now reached a with more automated deployment and life
VNFs from multi-tenants. The traffic good maturity level. This has been cycle management, you get all the layers of
running on these VNFs is typically around achieved by Ericsson customers with NFVI, VNFs and MANO to work together.
30 percent, ranging from 10–50 percent our introduction of pre-integrated and The benefits will continue to increase, as
depending on type of VNF. The uptake system-verified solutions. This more parts of the full end-to-end system
among leaders is increasing fast. could apply for all types of cloud work well together. Success is increased
We have seen Management and infrastructures regardless if they are by using standardized solutions such as
Orchestration (MANO) solutions evolve Virtual Infrastructure Manager (VIM) OpenStack, Kubernetes or CNCF, and
from basic VNF Manager (VNFM) to based NFVI or bare metal. Notable by actively contributing to the key open
more advanced, extended VNFM plus here are the recurring costs for life source projects to limit any forks and keep
cloud orchestration (NFVO) and we cycle management of the stack where the maintenance costs down.
also see early end-to-end service the open source software is upgraded
orchestration deployments. every quarter. This also enables solution
Communications service providers (CSPs) support of the entire cloud infrastructure
are starting to change their operations stack, through a single interface.
organizations and processes to manage One trend has been to move
horizontal cloud platforms with multiple from totally disaggregated NFVI
functions and applications. stacks to more pre-integrated and
All this has created a platform system-verified solutions from one
both in technology and learnings to vendor, although the solution may be
take on the next step of evolving the both open source and multi-vendor.
telco cloud infrastructure to run cloud It is key to limit complexity and strive
native applications. for pre-integrated configurations to

Figure 3: The NFV infrastructure is the foundation for 5G Core and 5G applications

Critical services and

Reliability/telco grade
infrastructure control

5G Core applications High performance

Sensor networks
and low latency

5G Core

Critical control of
Central management
remote devices NFVI

Smart vehicles, transport Distributed architecture

4 Ericsson  |  Building a cloud native infrastructure

5G implications in the
cloud infrastructure

When NFV started, before 5G was a concern, the focus was on

cost and vendor independence. 5G has changed the agenda.

With 5G, the idea is to build a single

infrastructure for all types of IoT and
enhanced mobile broadband services,
ranging from gaming or highly demanding
automated manufacturing, to low traffic
sensors. The solution to this is a shared
infrastructure with orchestrated network
slices to efficiently deploy all the services.
This now means that a working telco cloud
infrastructure is a prerequisite to run a lot
of the new and upcoming 5G services and
future innovative applications.
To have services set up using network
slices will require a service orchestrator The arrival of 5G is changing all industries, including the approach to cloud infrastructure
that can tie VNFs or cloud native network
functions (CNFs) together. Long term, Edge computing and distributed Management of the 5G
this can be automated and connected to cloud native infrastructure cloud infrastructure
business support systems. In the 5G context, edge computing Life cycle management will be more
This means that a competitive cloud has been discussed as a means to important as we move to cloud native
infrastructure for 5G services needs to provide capabilities like low latency and use more open source software. With
be stable and able to onboard VNFs and or device offloading for new services physical nodes, hardware upgrades are
CNFs from a broad ecosystem of suppliers, or dedicated enterprise networks. years apart, and software upgrades just
and these must be possible to orchestrate We foresee that the same type of telco once or twice per year. With continuous
based on exposure APIs. cloud, based on ETSI NFV, will also be integration and continuous delivery (CI/CD)
For the cloud infrastructure, this used in the evolution to cloud native. around the corner, we are foreseeing
means that it will be key to have a Most important, however, is that weekly or daily upgrades. This means that
horizontal NFVI platform capable of the entire distributed cloud can be management is quickly becoming one of the
running multi-tenant VNFs and CNFs, orchestrated by the same service most important factors to consider when
and to have those orchestrated. orchestration domain. implementing a cloud native infrastructure.

Figure 4: Automated and optimized workload placement across distributed

data centers in a multi-domain, multi-technology and multi-vendor environment

MANO and service orchestration

slices

App App CNF CNF VNF VNF

App App CNF VNF CNF CNF VNF VNF VNF CNF

CNF VNF CNF VNF CNF CNF VNF CNF

VNF VNF
PNF PNF
Caas Caas Caas
Iaas Iaas Iaas

Access cloud/Edge Central office Central data center Public cloud

~1–5ms RTT latency ~5–20ms RTT ~20–50ms RTT latency

Note: This figure refers to physical network functions (PNF), virtual network functions (VNF) and cloud native network functions (CNF)
5 Ericsson  |  Building a cloud native infrastructure

Evolution to
cloud native

Evolving network functions to cloud native isn’t

a one-step process; it is a multi-year journey.

A key learning from the virtualization Figure 5: Multi VNF NFVI infrastructure with CaaS platform
journey with our NFVI solution is that
system-verified solutions work best MANO
and are the most efficient, and that
customer-unique and best-of-breed VNF VNF CNF CNF CNF
NFVI stacks come with additional work,
costs and a longer time to market. Kubernetes (CaaS) CNI + CSI
The VNFs for some parts of the
ecosystem will continue to be in use for
several more years and those still need Virtualization (IaaS) Hypervisor + VM +SDN
to be supported.
We see two main scenarios for service
providers evolving the network towards
cloud native. Hardware management Compute Network Storage

Option 1: add a container

platform on the NFVI
For CSPs with existing NFVI platforms, an Option 2: deploy a new bare metal These solutions will start to come out during
easy way to start with cloud native-based infrastructure without VIM late 2020 for a limited set of CNFs, like the
applications or CNFs is to add a container An alternate deployment option is 5G Core, to some leading CSPs.
as a service (CaaS) platform on top of the to deploy CNFs over a bare metal In the early days it has fewer features
existing VIM (see Figure 5). infrastructure to leverage the full benefits than today’s NFVI, but this will pick up in
This means that the industry standard, of cloud native transformation. This the next few years.
CNCF-certified Kubernetes will be solution has no VIM and instead the Today, this solution will already be
deployed in virtual machines over an Kubernetes-based, CaaS platform runs attractive for CSPs who seek to move
Openstack-based infrastructure as a directly on the underlaying hardware. first towards a bare metal platform for
service (IaaS). We believe this will be the long-term their cloud native applications. It may
This is attractive for those with stable solution, as it simplifies the architecture also be an alternative for those that still
working horizontal cloud platforms with and uses the underlaying hardware more don’t have a unified virtualized core to
many VNFs. This solution offers: efficiently, with significant TCO savings. expand upon.
• Easy evolution from 5G Evolved
Packet Core to 5G Core on existing
NFVI virtual infrastructure
• Virtualized and cloud native in the Figure 6: CNFs on bare metal infrastructure
same infrastructure
• MANO orchestration fully working MANO
from day one
• NFV multi-tenancy CNF CNF CNF CNF CNF
• Multi-vendor VNFs and CNFs
• Security Kubernetes (CaaS) CNI + CSI
• Automated operation
• Simplified secondary networking handling

All in all, this will give a fast time to market Hardware management Compute Network Storage
for 5G enterprise and IoT services.
6 Ericsson  |  Building a cloud native infrastructure

Figure 7: Cloud native transformation journey and deployment options

Current Next Future

Cloud Cloud
VNFs Prioritizing benefit and VNFs + native Evolved MANO and VNFs + native
minimizing impact apps bare metal CaaS apps

MANO MANO MANO

Open stack API Open stack API K8s Helm API Open stack API K8s Helm API
CEE (IaaS) CCD (CaaS)
SDI CEE (IaaS) CEE (IaaS) CCD (CaaS)
SDI SDI

Cloud
native
apps

MANO
K8s Helm API
CCD (CaaS)
SDI

The cloud native infrastructure Long-term benefits of a bare • Application performance is better
transformation journey metal cloud infrastructure and more deterministic on bare metal
We can foresee a journey starting with Some may ask why we should move deployments since bottlenecks like
a smaller virtulization with few VNFs, to cloud native, as well as a bare metal the guest operating system and
which is expanded to a multi-VNF infrastructure, now that we have just virtual switches are removed with the
horizontal NFVI with MANO for got our OpenStack-based NFVI platforms virtualization layer.
multi-vendor VNFs. This platform is then to work. So, what are the long-term • Automated life cycle management and
expanded with a CaaS layer to support benefits of Kubernetes (CaaS) over the seamless introduction of CI/CD are
5G Core and other CNFs. Figure 7 is a bare metal infrastructure? crucial benefits of a bare metal cloud
schematic and will also apply for the • Network deployment and operations infrastructure. These functionalities
distributed cloud and edge computing. are simplified through easier life are challenging to introduce in a cloud
cycle management of the cloud stack with many dependencies, but
infrastructure. Since the virtualization with a simplified solution using fewer
layer is not part of the cloud stack, layers, automation will be easier to
fewer teams are involved, thereby achieve end-to-end.
streamlining management and • Another advantage is that the
operations. The reduced complexity introduction of new hardware
also results in faster fault isolation acceleration technologies, like smart
and troubleshooting. network interface cards and support for
• With the elimination of the graphics processing units, needed for very
virtualization layer, the infrastructure demanding applications become easier
overhead is drastically reduced. for new 5G use cases.
More compute and storage resources
are now available for application Open source is also
deployments thereby increasing important for bare metal
hardware efficiency. This is especially With dedicated teams of top developers,
important for edge computing, which we are actively making upstream
often have resource constraints in contributions in some of the key telco
remote sites. relevant open source projects, such as
• Another big advantage is that no Metal3-io, cluster-API, Redfish, kubeadm
software license fees are needed for and Kubernetes IPv6 dual stack. These are
the virtualization software, resulting in critical for making an industrialized, telco
substantial TCO savings for CSPs. grade and production-ready, bare metal
CaaS infrastructure.
7 Ericsson  |  Building a cloud native infrastructure

5G Core and cloud native open up new, potentially revenue-generating use cases

Figure 8: Ericsson’s end-to-end solution support Summary

The introduction of 5G Core opens up the
Customer potential for new revenue-generating
services in the IoT and enterprise spaces,
Single point of contact based on network slices. This imposes a
paradigm shift as the core network
now needs to support CNFs and
Ericsson Dynamic Orchestration container-based software.
This is done by introducing a
VNF VNF CNF CNF CNCF certified CaaS platform based
E2E support

E VNFM
on Kubernetes.
Key characteristics include efficient
IaaS VIM CaaS, Cloud Container Distribution automation of CI/CD loops and automated
cloud execution environment life cycle management operations of
CNFs over a managed software-defined
SDI infrastructure (SDI).
Central sites Central sites We see two alternatives for the
evolution of the cloud infrastructure:
• The existing NFVI is complemented with a
3PP Ericsson 3PP
CaaS platform. This allows for a possibility
to run VM-based applications in parallel.
This will be beneficial if the current NFVI
is a stable multi-VNF telco cloud.
Solution support for cloud infrastructure isolation, issue and emergency recovery • A new bare metal infrastructure without
Multi-vendor and multi-layer virtualized services, root cause analysis, management a virtualization layer is introduced
stacks always come with the question of of the integrity of customers’ solutions and alongside the current virtualization
who takes the end-to-end responsibility; vendor coordination. platform or platforms. This solution will
who invests in the competence needed to Enhanced by automation, engineered have a simpler stack and make better use
support the solution and secure stability intelligence assets, and reproduction of the underlaying hardware. It enables
and performance? across the multi-layer solution levels more efficient CI/CD capabilities, and
A system-verified cloud stack brings using reference test configurations, a will provide the long-term type of
the possibility to provide CSPs with a combination of local and proven global architecture for the telco cloud. This will
single point of accountability for solution expertise in comprehensive customer start to be introduced during 2021 in
support tasks, including ecosystem fault support services can be offered. some operators’ networks.
 Ericsson enables communications service providers
to capture the full value of connectivity. The company’s
portfolio spans Networks, Digital Services, Managed
Services, and Emerging Business and is designed to
help our customers go digital, increase efficiency and
find new revenue streams. Ericsson’s investments in
innovation have delivered the benefits of telephony
and mobile broadband to billions of people around
the world. The Ericsson stock is listed on Nasdaq
Stockholm and on Nasdaq New York.

www.ericsson.com

Ericsson The content of this document is subject to 3/287 01-FGM 101 51

SE-164 80 Stockholm, Sweden revision without notice due to continued © Ericsson 2020
Telephone +46 10 719 0000 progress in methodology, design and
www.ericsson.com manufacturing. Ericsson shall have no
liability for any error or damage of any kind
resulting from the use of this document