Network Transformation and 5G

Looking beyond the speed and latency promises

It’s inescapable that 2019 is the year for 5G. The big promises of high bandwidth and low latency are now accompanied by the first phase of commercial network deployments. But a large part of the supporting story doesn’t get the same level of attention, despite its central role in delivering the real value and promise of 5G.

CCS Insight’s experiences on several commercial 5G networks have demonstrated the advantages in downlink speed and capacity that are an important first step in the story (see A Swiss 5G Adventure). However, progress in the radio access network (RAN) is inseparable from the need to update the core network. The majority of the 5G networks launched in 2019 will be non-standalone, meaning they use the existing LTE RAN and core network. This was a logical step to get services to market quicker and avoid fragmentation through proprietary implementations.

However, it also shines a light on the bigger opportunity for 5G. For the technology to really deliver its promise it must be accompanied by broader transformation of the RAN and core network. In the near term, lower latency, better uplink performance and greater efficiency will all require a shift to standalone networks and a dedicated 5G RAN. More importantly, the flexibility of a software defined network is needed if 5G is to deliver the service value to opportunities ranging from large-scale implementations of the Internet of things to mission-critical services.

Ericsson recently published its annual Mobility Report indicating 82% growth in global mobile data traffic between 1Q18 and 1Q19. With this massive increase in data consumption, the network must better understand usage and anticipate demands. This is central to delivering a high quality of service and increasing network efficiency.

This is what’s driving transformation as the network moves from a complex array of proprietary, fixed-function devices to one more akin to a data centre, where more general-purpose servers can run virtualized network functions using open-source architectures. The hardware is abstracted away though software to decrease cost and increase agility and efficiency.

This represents a significant opportunity for a raft of companies, not least the infrastructure players. What is often overlooked is the role Intel, Cisco and others play in this equation as 5G becomes defined as much by computing as communications. CCS Insight believes that Intel stands to become part of the fabric of the network in the same way that it’s the backbone of the data centre.

The network transformation opportunity can essentially be split into three parts. The first is software defined networking, in which the network becomes fully programmable using switches configured through a protocol such as OpenFlow or other proprietary platforms.

The second element is network function virtualization — network functions are run in containers or virtual machines using open-source software in much the same way as a data centre operates. This means functions are programmed in software, decoupled from the underlying hardware and can be easily connected. The result is better optimization and easy redistribution depending on the needs of the network or applications.

The third part, mobile edge computing, benefits from this transformation of the RAN and core network by enabling critical functions and applications to be run closer to the user and source of the data. However, although mobile edge computing is benefiting from significant hype and is touted as a key part of 5G, it’s dependent on the transformation of the broader network.

This represents a significant opportunity, not least because network operators are still in the early stages of this transition. AT&T hopes that by 2020 three-quarters of its core network will be virtualized. Verizon has said its core network will be fully virtualized within two years and Rakuten has worked extensively with Intel to become the first 100% cloud-native mobile network when it launches later in 2019.

Indeed, Rakuten illustrates the opportunity for a wider number of participants as the rigid supply chain that characterized fixed-function devices is displaced by open-source approaches to virtualized network functions. Rakuten’s network uses Xeon processors and field-programmable gate arrays from Intel, servers from Quanta alongside other providers such as Altiostar and Cisco.

Such levels of flexibility and programmability are very new concepts for most network operators, and they will need time to establish themselves, along with investments in tooling and support. This is comparable to the transformation that computing has gone through with the cloud. Many of the protagonists in hyperscale data centres will be prominent in the “cloudification” of the network and the process won’t just transform mobile, it’ll affect the cloud itself.

This article was first published by FierceWireless on 22 July 2019.