Operator ‘Network-as-a-Service’ Hopes Requires a Re-Think of the User Plane Function

The User Plane Function (UPF) is embedded within 5G networks and coordinates decapsulates data traffic between the Radio Access Network (RAN) and the data network (internet). Simply put, it ensures that 5G networks data packets are quickly and accurately routed to the correct destination on the data network. Beyond packet routing and forwarding, the UPF offers other crucial functions, such as quality of service and packet inspection. As the interconnection between 5G networks and the data network it provides grows, so does the aggregation and processing capabilities to deploy enterprise-specific edge services and run them over the telecom network (this is its true end-user value). Most operators currently run UPFs in the core network, though since the inception of 5G, stakeholders have discussed pushing deployment towards the edge to be closer to end-users and applications.

Pushing the UPF towards the edge may not be a new concept but implementation is yet to widely follow. However, last year the market showed renewed interest in the commercial value of UPFs in distributed compute deployment. Equinix announced ‘the 5G Edge Developers Framework on Equinix’; this framework provides a simple tool for operators to deploy UPFs in real-time with full automation across Equinix Fabric locations. By placing UPFs in major metro areas, close to end users, cloud, and internet exchanges, it has shown operators improvement in enterprise application experience, subsequently enhancing their 5G network Business-to-Business (B2B) commercial opportunities.

More recently, the Mobile World Congress (MWC) of 2023 was buzzing with news and conversations around UPFs. NTT Docomo and NEC Corporation announced successful on-boarding of NEC’s 5G user plane function to edge computing on AWS Graviton3 Processor. This step forward is meant to improve 5G network performance and give greater control over network energy consumption, while supporting scalable application workloads across distributed compute environments. In addition, Intel announced its fourth Gen Xeon Scalable processors which demonstrated throughput of 1 terabit per second in for 5G User Plane Function (UPF) as part of their Converged Edge Media Platform.

The market is finally moving from discussion around the location of UPFs towards development and implementation. Necessity for 5G network optimization and commercialization seem to be at the root of this trend:

• Support B2B service growth to support 5G monetization: Deploying high value enterprise-ready applications at the network edge is essential for operator revenue generation. UPFs, deployed in the right location and with optimal throughput, can help lower network latency providing the foundation for millisecond use cases. Legacy UPFs will just slow data transfer and impede application performance, lowering the likelihood that enterprises will adopt 5G edge services.
• Required optimization of energy consumption for Total Cost of Operation (TCO) management: Energy prices are rising and the dissemination of data creating devices is driving up network data, meaning networks are costing more. Updating UPFs can introduce traffic offloading to lower network traffic, activate sleep mode to reduce consumption when not processing traffic, and packet batching to increase the number of data batches processed at once.
• Support networking slice delivery: Often viewed as a substitute to private networks, network slicing is one of the most anticipated services that can be delivered through 5G standalone. The UPF enforces network slice’s Quality of Service (QoS) and alignment with use cases. Given expectations that hundreds of slices should be deployed and rolled out regularly, the UPF’s traffic management role is vital. It is also important to re-think the position of the UPF, as deployment in the core will be less responsive to required changes in network slice deployment.
• Deploy automation capabilities through Application Programming Interface (API) programmable: The UPF can support network automation deployment, which has massive ramifications for network operations as by adding intelligence, manual operations can be reduced by enabling network self-optimization and dynamic resource allocation.
• Keep the network in step with API developments: Typically, operator networks are static. However, with the industry renewed interest in APIs, operators must respond by making the network more agile, customizable, and able to support enterprise network applications. The UPF plays an important role in the optimization of data traffic and lowering latency between end-users and network services. It can help give the operator more granular control over network connection deployment, as well as enabling edge applications, these are key steps needed to support a transition closer to ‘Network-as-a-Service’ deployments.

Supporting UPF development will have a positive impact on 5G revenue generation and cost optimization, but what exactly should operators do?

• Use emerging edge developer platforms to ease virtualized UPF deployment: Partnering with digital infrastructure providers (like Equinix) will streamline this process and support containerization through a mature developer platform. Developer platforms are becoming increasingly important tools as the edge becomes more popular for operator function deployment. Developers want simplicity (through low/no code deployment), low time-to-value, and single pane of glass visibility across hybrid and edge environments. Platforms like 5G Edge Developer Framework on Equinix and Azure Operator Nexus should a pillar of any operator strategy as they look to unlock the value of deploying network functions at the edge.
• Move UPF deployment from core to edge: Most operators rely on the core to run UPFs, however, as applications move closer to the edge, this will impede latency and performance. Although it will be costly to move the UPF outwards, by increasing its proximity to the end-users and network edge applications, data traffic can be more effectively handled. This will have significant performance advantages for edge applications and help operators position themselves as leaders in the edge service market. If we consider Network-as-a-Service (NaaS), controlling data traffic from the edge is essential to deploying many high values network services, including bandwidth-on-demand and network slicing.
• Engage with partners to support UPF distribution: Distributing UPFs closer to the end-user will be costly in terms of operations (deployment, management, upgrades, etc.) and infrastructure. Third party infrastructure vendors, like neutral hosts and hyperscalers, could lower the cost burden on operators by providing scalable edge infrastructure and infrastructure management services. In addition, neutral host infrastructure sharing will lower deployment costs and support a more realistic business case for distributed UPF deployment.

• Automate UPF activation and provisioning: Deploying user plane functions at the edge across the network is time consuming. Automation of activation, provisioning, and orchestration is a gamechanger that can ensure that new UPF deployment model is repeatable and does not bring operational chaos or higher costs.

Telecom digital transformation, and eventual NaaS deployment, is an arduous process involving significant structural and technological investment. For ABI Research, distributing the UPF is a vital step. It provides greater control over the network improving performance for vertical-specific edge application delivery and the deployment of NaaS (bandwidth-on-demand/network slicing). Renewed industry focus on 5G UPFs shows operator commitment to pursue an aggressive B2B 5G commercialization strategy. Hopefully, this intent will lead to tangible progress, as operators sorely need big 5G commercial wins and need them soon.