Networking for Nerds

Unlock the True Potential of 5G With Edge Infrastructure

Learn why deploying the UPF in the right location matters, and how you can get the edge infrastructure you need to do it

Rodney Elder
Oleg Berzin
William Lloyd
Unlock the True Potential of 5G With Edge Infrastructure

Global 5G adoption is now well underway, with almost 1.7 billion 5G devices already in use. This number is forecast to continue growing rapidly over the next several years, reaching almost 5 billion by 2028.[1] Looking forward to this new era of growth, the question remains: Are we prepared to do 5G in a way that truly lives up to the hype we’ve heard in recent years?

The entire premise of 5G is that it can enable enterprise-grade wireless services, where previous generations of wireless technology were only capable of supporting consumer services. It’s still not clear whether we have the network infrastructure we need to do that on a global scale. In order to use those billions of 5G devices to their full potential, we’ll need new and optimized 5G networks to support them.

Optimized 5G infrastructure must start with the UPF

For a closer look at why our current 5G infrastructure may be falling short, let’s consider the user plane function (UPF). The UPF is the hub for all critical interconnection that takes place on 5G networks. It sits between 5G users and the applications they want to access.

Create digital advantage at software speed

Learn how to adopt a digital-first strategy to support enterprise transformation and expand your capabilities

Download Whitepaper
Light speed explosion illustration long exposure photography

The UPF takes encapsulated 5G user traffic originating from the radio access network (RAN) and de-encapsulates it so that it can travel over traditional networking infrastructure. This allows 5G traffic to flow from the UPF to outside networks such as internet, edge computing and cloud ecosystems. Applications reside on those outside networks, which means that ensuring low-latency and predictable connectivity from the UPF is essential.

The UPF is managed by the mobile core, which performs control plane network functions such as the Access and Mobility Function (AMF) and the Session Management Function (SMF).

The UPF also supports intelligent network slicing and allows mobile or network service providers (NSPs) to route and prioritize certain subsets of traffic based on user need. This gives them a valuable way to differentiate their 5G business model by offering multiple classes of services for users.

Since the UPF is so important to enabling key 5G capabilities, where you deploy it matters. We think the industry as a whole has yet to fully recognize why deploying UPFs in the right location matters so much. Today, most telco operators are still deploying UPFs inside their core networks. These core networks worked well enough for their legacy data and voice services, but they simply aren’t as fit to support the latest 5G offerings. This is primarily because they lack proximity to users and applications, which can result in higher latency. Without reliable low-latency connectivity, the promised benefits of 5G performance start to break down.

At Equinix, we firmly believe that the UPF should be deployed at the digital edge to ensure low latency and operational efficiency in 5G networks.

The future of 5G will happen at the edge

Due to the key role it plays in handling traffic flow and enabling intelligent network slicing, the UPF exerts a powerful pull on the 5G traffic around it. This is why we like to say that “wherever the UPF goes, that’s where the traffic flows.” Deploying UPFs at the edge makes sense from a latency perspective, but it’s also important for network operators to recognize that 5G traffic will follow the UPF to the edge. This will cause increased demand for compute capacity in those edge locations. For this reason, it will become more important than ever for network operators to be able to stand up new edge infrastructure quickly and flexibly.

With this in mind, it’s no wonder that the Global Interconnection Index (GXI) 2023, a market study conducted by Equinix, found that service providers are growing their edge infrastructure 50% faster than their core infrastructure. Digital infrastructure partners such as Equinix can help NSPs get the edge infrastructure they need to expand their 5G service offerings in key metros around the globe.

NaaS will enable 5G at the edge

A key aspect of digital transformation for NSPs—one that goes hand-in-hand with the shift to the edge—is the growth of Network as a Service (NaaS) operating models. Forward-thinking NSPs recognize that replacing their traditional physical networking hardware with a NaaS approach can help them deploy network services quicker and more cost-effectively. Making this transition will help them increase network agility and manage resources more efficiently using an OPEX-based cost model. In short, they’ll be better positioned to lead in the digital-first era.

Looking at things specifically from a 5G perspective, adopting a NaaS operating model will help NSPs scale down their UPF deployments. This is important because today’s UPFs are simply too large to be effective. Major telco companies often have to dedicate multiple racks of server space to support a single UPF deployment. As 5G adoption picks up, these large UPFs will become increasingly untenable. In addition to being prohibitively expensive, they’re also much more difficult to deploy in edge locations.

By partnering with Equinix, NSPs can deploy their UPFs and other 5G functions as containerized cloud-native components. These virtualized functions represent the next evolution of 5G. They are quick and repeatable to deploy, manage and scale, thus helping NSPs get the 5G functions they need, when and where they need them. Since they can be deployed on a smaller footprint, they can provide the foundation for a thriving 5G traffic aggregation point without placing unnecessary demand on the available edge infrastructure.

Open-source implementations help developers test applications in an end-to-end 5G environment

To help mobile developers and architects prepare for the emerging future of 5G, Equinix is demonstrating a 5G Edge Development Framework. This new end-to-end solution is powered by free5GC, an open-source software package that brings together 5G functions—the UPF, the mobile core and the simulated UE/gNB—along with a virtual router to move traffic between them and connect to the internet and multiple clouds.

All 5G functions are deployed as cloud-native applications using Kubernetes inside virtual machines, while the router comes from our Network Edge portfolio of virtual network functions (VNFs). This means the framework can be spun up on demand in any of the 28 global metros where Network Edge devices are currently available, with no physical networking hardware required.

Together, the four elements of the solution allow mobile developers to test and run applications and frameworks in an end-to-end 5G environment with a simulated user edge and multicloud-hosted applications. The solution includes all the advanced features and functions of a real-world 5G Stand Alone (SA) environment, so developers can experience how their applications would perform in a real-world scenario.

5G Optimized by Real-Time Edge (FORTE)

The 5G Edge Developers Framework is a first step toward making it quicker, easier and more cost-effective for Equinix customers to deploy UPFs in edge locations, and thus unlock the full business potential of 5G on a global scale. With Equinix digital services like Equinix Fabric®, Network Edge and Equinix Metal®, customers can deploy optimized 5G infrastructure repeatedly using APIs and Infrastructure as Code (IaC) tools such as Terraform.

In addition to the automated provisioning, developers and architects deploying the free5GC package on Network Edge can quickly see the benefits of placing production UPF solutions in major metros close to public clouds and internet exchanges by observing the latency on the N6 interface of the UFP (the internet/cloud-facing interface). Users typically experience round-trip time (RTT) latency under 5 milliseconds.

Visit to learn more about the package, community and support options.

The slicing must go on

In general, network slicing refers to the ability to provision and connect functions within a common physical network to provide the resources needed to deliver services under specific performance constraints (such as latency, throughput, capacity and reliability) and functional constraints (such as security and applications/services). It’s important to note that the word “network” here refers to the complete system that provides services, and not specifically to the transport and network layer functions that are part of this system. While network slicing is typically considered a feature of the mobile network, we can make the following observations about how it works across the complete system.

The mobile traffic originates in the mobile network, but it is not contained in the mobile network domain. Therefore, to preserve intended characteristics, the slice created in the mobile domain must be extended all the way to where the traffic wants to go. This leads to the concept of “the slicing must go on.”

The placement of network functions within the slice must be optimal, relative to the slice performance requirements, so that users can ensure the intended performance end-to-end. One of the critical network functions of a slice is the UPF. For this reason, NSPs must place or activate the UPF in optimal locations relative to the end-to-end user plane traffic flow.

We expect that private hybrid multicloud connectivity will remain a key requirement for enterprises when they use 5G. In this case, the “hybrid” part refers to the private edge computing resources (we also loosely call it MEC) located in Equinix metros, while “multicloud” refers to the need to access multiple cloud providers from 5G devices. To ensure both elements of hybrid multicloud connectivity, NSPs must integrate the UPF with the interconnection fabric.

Because a slice must span multiple domains, it’s also critical for NSPs to automate UPF activation, provisioning and virtual interconnection to edge compute and multicloud environments.

With Equinix Fabric, our software-defined interconnection solution, NSPs can move 5G traffic from the UPF to outside networks while still preserving the properties of the various network slices. This not only helps ensure low latency, but also allows NSPs to perform network slicing on an end-to-end basis. The ability to optimize certain subsets of traffic throughout the entire 5G traffic lifecycle can open up important new business opportunities for network operators.

Check out the white paper Create a digital advantage at software speed for a closer look at how solution providers can cost-effectively meet enterprise customer demand for advanced technologies like 5G, edge computing and AI.


[1] Ericsson Mobility Visualizer


Rodney Elder Former Global Principal
Avatar photo
Oleg Berzin Senior Distinguished Engineer, Technology and Architecture, Office of the CTO
Avatar photo
William Lloyd Corporate Development Director
Subscribe to the Equinix Blog