Skip navigation
Like what you’re reading?

Cloud-native telecom applications – a good fit for Kubernetes on bare-metal?

Available in English Français

Is there a reason to develop a truly cloud-native infrastructure? We believe there is, particularly when considering the advantages of running cloud-native network functions directly on bare-metal hardware, without the requirement of a virtualization layer. To explore this captivating topic in more detail we've released a new edition of the cloud-native 5G Core network guide series dedicated to Kubernetes over bare-metal.

Product Marketing Manager

Person looking over a lake

Product Marketing Manager

Product Marketing Manager

A core network for 5G is more than 5G Core: One of the fundamental domains is the infrastructure part, where we’ve released a new edition of the guide to deploying Kubernetes over bare-metal infrastructure for 5G

 

Guide to deploying Kubernetes over bare-metal infrastructure for 5G

Figure 1: A guide to deploying Kubernetes over bare-metal infrastructure for 5G

 

We wanted to expose the impact of adopting cloud-native and the new 5G Core architecture has on telecom networks. With that as a backdrop, the cloud-native infrastructure guide is part of a greater package of a guide series covering five other domains in the network: 5G Core, 5G Voice, automation and orchestration, operations and life cycle management, and security. A new edition of the entire series is available for download.

 

Monetization of 5G is top of mind

As most people would agree, 5G not only enhances the user experience for consumer mobile broadband but also offers a multitude of other benefits. With the rollout of 5G networks also comes the ability for communication service providers (CSP) to address new business opportunities spanning multiple industries and enterprises. Top areas are the manufacturing and automotive industries, private and mission-critical networks, and fixed wireless access.

In a relatively short period of time our industry has undergone three stages of technology: from traditional native nodes (hardware + software + services integrated) to virtualized networks (decoupled hardware from software/services), to the current cloud-native networks (a service-based architecture using microservices). 5G Core  is cloud-native, which represents a paradigm shift that offers many opportunities, but also challenges organizations to fine-tune their operations and adapt their processes. Monetization of 5G is top of mind for our industry, and 5G is use-case driven. The number of use-cases in the pipe is plentiful and we see interesting new cases coming to life, not the least from CSPs that are getting production ready with their offerings for enterprise solutions based on network slicing. This is only the beginning – more cases are to be expected with the innovation that's anticipated to be happening on top of networks.

 

More capabilities

The new 5G Core architecture is more dynamic than its predecessors, meaning that new and innovative services can be launched more quickly to test market demand, and be phased out if not successful.

Much of this agility stems from the fact that the 5G Core architecture is cloud native. It also means that the new 5G standalone network comes with new and enhanced capabilities for low latency, high bandwidth, network slicing and programmable networks.

The proposition is to build a single infrastructure for all types of IoT and enhanced mobile broadband services – from gaming or demanding automated manufacturing to low-traffic sensors. The solution is a shared infrastructure with orchestrated network slices to efficiently deploy all the services.

Our industry is evolving at an ever-accelerating pace to the new, standardized 5G Core based on cloud-native design principles. From a cloud infrastructure perspective, the complexity grows in two dimensions:

  • The number of layers and functions in the stacks of hardware, virtualization layer, applications, and operations & business support systems
  • The number of potential combinations within and between the layers with products from many vendors’ open-source software

 

How to handle growing complexity

The cloud-native infrastructure guide elaborates on how to handle this new level of complexity, as well as the central question how our industry can increase efficiency and lower the opex and capex when data traffic keeps growing? Concluding that cloud-native telecom applications require cloud-native infrastructure, the answer boils down to a reasoning around some key considerations:

  • The importance of having a horizontal platform that’s capable of running multi-tenant Virtualized Network Functions (VNFs) and cloud native network functions (CNFs), and to have those orchestrated
  • That the infrastructure operations need to be unified and centralized, including end-to-end management of the cloud infrastructure
  • The need for a high level of automation, covering everything from handling open-source software that’s subject to frequent upgrades, to setting up services using network slices connected to the business support system

Cloud-native brings the opportunity to manage and hide the complexity of networks and simplify operations, which plays a critical role when launching new 5G services.

With cloud-native, along with continuous integration and continuous deployment (CI/CD), software pipelines, etc., our industry is equipped with the technologies, processes, speed, and agility needed to deploy new services and rapidly give life to new use-cases.

 

Two infrastructure implementations to consider

To provide support for cloud-native telecom applications, CSPs have two alternative infrastructure options to consider. One option is to introduce a Kubernetes-based container as a service (CaaS) platform that runs directly on the underlying hardware, and the other option is to deploy Kubernetes in virtual machines (VMs) over an OpenStack-based infrastructure as a service (IaaS) platform.

The cloud-native infrastructure guide discusses pros and cons for both these alternatives.

The conclusion is that to provide support for cloud-native applications in the fastest and most cost-efficient way, it’s recommended to target the Kubernetes over bare-metal cloud infrastructure option. For many of our service provider customers, it’s proven clear that this is the most competitive solution for cloud-native applications as it simplifies the architecture and uses the underlying hardware more efficiently, which in turn translates to total cost of ownership (TCO) savings.

The other option, to add a CaaS platform on top of the existing virtual machines, should not be ruled out for CSPs with an existing network functions virtualization infrastructure (NFVI) platform. This solution could be attractive for networks with stable working horizontal cloud platforms with many VNFs in commercial operation.

An example of a customer that’s chosen the Kubernetes over bare-metal path is SK Telecom in South Korea, which became the first service provider in the world to go live with Ericsson 5G Core deployed on Ericsson Cloud Native Infrastructure solution.

 

Why move to Kubernetes over bare metal?

Why would a service provider move to a bare-metal infrastructure if there’s already a well-performing OpenStack-based NFVI platform in place?

The cloud-native infrastructure guide makes a note that this is a common question in the telecom industry, with answers summarized:

  • Automation of a Kubernetes over bare-metal cloud infrastructure will be easier to achieve since the lack of a virtualization layer significantly simplifies the architecture, with fewer products and components to maintain, and fewer people needed to operate the technology
  • A Kubernetes over bare-metal infrastructure is more suitable for distributed cloud infrastructure deployments to provide edge solutions, where 5G Core or Cloud RAN applications are needed
  • The introduction of new acceleration technologies (smart NIC, etc.,) becomes easier for new 5G use cases.
  • Application performance is better and more deterministic
  • Savings in the magnitude of 30 percent are possible, including costs for energy, operations and maintenance, software licenses, and hardware

 

Final take-away

The introduction of cloud-native applications imposes a paradigm shift for CSPs.

A Kubernetes over bare-metal infrastructure delivers a simplified architecture compared to running containers in VMs, and greater efficiency, performance, and automation can be achieved.

A Kubernetes over bare-metal solution has a simpler stack and makes better use of the underlying hardware compared to virtualized infrastructure, enabling more efficient continuous integration and continuous deployment capabilities.

 

Read more

Download your guide: Building a cloud-native infrastructure

Read more in our cloud-native 5G Core network guide series 

Read more about why cloud native infrastructure based on bare metal is the future 

Read more about 5G Core 

The Ericsson Blog

Like what you’re reading? Please sign up for email updates on your favorite topics.

Subscribe now

At the Ericsson Blog, we provide insight to make complex ideas on technology, innovation and business simple.