The Journey to a Cloud-Native Core Network

Cloud scale and flexibility a prerequisite for new services and revenue

I recently wrote about the evolution of the radio access network (RAN) as it becomes increasingly virtualized and cloud-like in order to fulfil the promise of 5G (see Virtual RAN: Separating Fact from Fiction). In this follow-up, I look at the parallel transformation of the core network. The “cloudification” of the core network is well underway and is perhaps the most important part of the equation, as it contains all network controls, manages authentication, the flow of data and essentially functions as the brain or control center of the network.

AT&T is projecting to control 75% of its core network functions with software by the end of 2020. At the end of 2019 Verizon said its journey is 60% complete.

This process of moving the core network elements away from fixed-function hardware and into software running on standardized hardware such as Intel Xeon processors isn’t new. The transition was sparked by 4G with network function virtualization that integrated voice and data and started the process of separating control and user planes. Several companies played a big role in this evolution of the network, including Intel, which helped mobilize an ecosystem toward the use of standards-based platforms and introduced efforts like the Data Plane Development Kit. This has become a de facto standard for running packet processing on general-purpose servers. The continued evolution of the network today builds on this foundation from the last decade.

The arrival of 5G and operators’ desire to deploy it quickly saw networks largely using non-standalone infrastructure, meaning that the 5G air interface was controlled by an operator’s existing 4G core network.

I expect this to change in the second half of 2020 and beyond. The need to generate revenue from 5G through new services and create more flexibility to offer low-latency uses at the network edge calls for further transformation of the network. As CCS Insight has highlighted before, if 5G is to achieve its promise and deliver rich capabilities, a high-capacity and low-latency air interface isn’t enough: it must be supported by core and RAN architectures with the flexibility and agility to match.

This vision requires a move to a cloud-native ethos, in which network functions are designed for maximum efficiency, speed, scale and capacity, based on established cloud infrastructure. There are three characteristics to this approach: the network gets further disaggregated into discrete functions, it becomes highly automated, and functions or applications can be easily composed using containers.

The transformation of 4G core networks was built on the principle of functions running in virtual machines; the next phase will shift to containers running functions as microservices. This enables code to be reused for different functions, fast creation of new services, the connection of multiple containers and seamless upgrades of individual functions.

In other words, the network becomes cloud-native. It’s designed, managed and scaled in the same way as the cloud. This has huge ramifications, not all of which can be covered here, but they include new services, new sources of revenue, the growth of open-source systems, such as Kubernetes, changes to hardware requirements and new participants.

Microsoft has been particularly active in targeting opportunities in the mobile network by acquiring Affirmed Networks and Metaswitch. However, others are also sharpening their ability to extend cloud services to the network edge. Over the past 12 months Google has unveiled its Global Mobile Edge strategy and partnered with AT&T, and Amazon has announced AWS Outposts, Local Zones and Wavelength (see Hyperscalers Accelerate the 5G Edge and Amazon’s AI Leadership Advanced at re:invent 2019).

It’s important to recognize the complexity of the transition to a cloud-native network. Operators won’t upgrade in a coordinated fashion, as their appetite depends on factors such as spectrum availability, 5G strategy, plans for monetization from services and the schedule for infrastructure investment. No two operators are the same. For those focused more heavily on mobile broadband in the near term, the need for a 5G cloud-native core is less urgent.

Furthermore, beyond a few greenfield exceptions such as Rakuten, most operators have a legacy network to manage and an existing 4G core network. Although the requirement to evolve is clear, operators can’t ignore the need to make the most of their investments and manage their current networks. Operators will almost certainly choose to run a 5G core network alongside a 4G core for a given period.

This highlights the complexity of the task facing operators today. They need to align business priorities, capital expenditure and existing assets with the need to modernize and embrace a new cloud-based architecture. Not all have the in-house capability to navigate the transition. This alone is a significant change, but operators also have to deal with other big transitions happening in parallel: a new 5G air interface, the need for transformation of both the RAN and the core network, the shift to multi-access edge computing and a significant yet uncertain opportunity in services.

The need for partners that understand this challenge and can accommodate varying business and technical requirements is essential. As the network becomes more cloud-like, it needs technology partners that can provide flexible and scalable solutions across the entirety of the network to fulfill a wide range of requirements. We will see a broad mix of legacy players and new entrants that will need to work collaboratively to deliver this cloud-native transformation.

A version of this blog post was first published by FierceWireless on 3 August 2020.