Your browser is not supported. Please update it.

30 September 2019

Telefónica will deploy a multivendor 5G core, despite complexity

Enrique Blanco, Telefónica’s CTO, was one of the eminent operator executives who warned, in 2017, that fast-tracking a non-standalone (NSA) variant of the 5G New Radio would deprive operators of many of the commercial benefits of 5G in the first few years, because most of those benefits require a 5G core. He remains impatient to move towards the full 5G architecture, even though his employer has not yet switched on commercial 5G networks even with NSA. But he is determined to deploy a multivendor 5G core, even though that will add to the already daunting complexities of the migration.

At the 5G Core Summit in Madrid this week, he said it would be “easier and cheaper” to work with just one vendor, but this would increase many risks, including those related to security and to potential outages. “It is easier to do with a single vendor in a single core, but we are doing a different approach to guarantee there is not any problem that can go viral and affect all customers. The industry needs to help us do this.”

To support guaranteed service availability, he plans to deploy a multi-layered core, with each layer supporting a different service domain, and potentially a different region. Different suppliers could be engaged to provide the functions for one or more layers and/or regions.

This very complex approach will either require a huge and costly investment in integration (inhouse teams or via a partner), or it needs open platforms to evolve quickly so Telefónica can stitch together its core using standard interfaces supported by all suppliers. So far, most efforts to achieve a multivendor core have been very individual to the operator – SK Telecom and NTT Docomo, for instance, are using a mixture of open and inhouse interfaces and APIs, which they require any supplier of virtual network functions (VNFs) to support, in order to deploy VNFs from more than one vendor in the core. But they acknowledge they rely heavily on integration and developer skills to optimize the platforms and make them carrier-grade – something BT has also discussed with regard to OpenStack-based deployments.

Blanco does not believe there is much time for the industry to get behind emerging open source and cloud-native technologies, if he is to be able to use these to ease the migration burden. His roadmap include full separation of the control and user planes, and a three-year transition to a microservices architecture.

At least the prolonged wait to move to full 5G may enable MNOs to avoid having to upgrade their cores twice within a few years. Had all 5G launches been standalone, the core would probably have been virtualized but not cloud-native, involving the MNO in two big migrations – first to virtual machines and then to containers a few years later. However, Blanco is one of the European CTOs who has insisted that the NSA compromise has forced operators, in order not to be left out of the early 5G race, to launch networks with limited performance impact.

When the Release 15 standards were first split between NSA and SA, he said this had removed many of the benefits of 5G, resulting in a first-stage platform with limited differentiation from 4G.

“My personal view is those operators who are really pushing for this – in the US, and Japan, and Korea – have different, more urgent priorities,” he said at the time. “Because of the Olympics, and whatever else, they want to deploy services using NR much sooner – which means they need the standards to be defined sooner, and the chipset makers to deliver sooner.” But he added: “5G is not just about sub-millisecond latency and 1Gbps speeds – we can get close to that with 4G already. So what’s new? The business model needs to be defined; 5G shouldn’t just bring new radios and antenna systems but new network architecture.”

Blanco has spoken before of the real importance of 5G lying in fundamental architecture changes, enabled by virtualization and software-defined networking, which will, in turn support new business models based on slicing and full agility. But none of this transformative capability will happen with non-standalone 5G NR, and he thinks 5G roll-out will take place over a longer time period than many expect, not a shorter one, because many of its key enablers, such as Massive MIMO, are being deployed first in existing 4G networks.

Orange’s director of technical strategy, Yves Bellego, has similar views. He has said: “5G is not and should not be limited to NR. We really believe we need a full 5G system… and there is a risk that defining a new radio and not addressing other portions does not in the end deliver the best solution. Having new radio does not answer all the needs.”