Mobile operators’ move towards virtualized, and even cloud-native, networks has been more aspiration than reality for most players until very recently. A burst of deployments of virtualized LTE packet cores in the 2010s largely proved to be the one swallow that does not make a summer.
Early virtualization platforms, including the telco-driven ETSI NFV (network functions virtualization), have proved to be very much a first generation technology. Far superior deployability and flexibility will come from cloud-native systems, and those will drive better economics in the 5G era – but have left would-be early adopters back in the phase of dealing with immature solutions that often come with risk or high upfront cost.
While the core is driving development of fully commercial solutions, and therefore operator uptake beyond a few pioneers, the RAN has remained firmly in the aspirational box. Even fervent adopters of network virtualization, such as AT&T, have put vRAN at the end of their roadmap.
This is for various reasons.
- The macro RAN is the most expensive part of an MNO’s spending, so return on existing investments in conventional equipment must be maximized, and migration to a new architecture must be de-risked as much as possible.
- This is difficult because the 5G macro RAN will make extraordinary demands on the cloud infrastructure, particularly the processors, that underpin it. The amount of processing power to support 5G beamforming or near-real time dynamic spectrum allocation is 10 times or more what LTE needed.
- Therefore, most MNOs are not yet convinced – sufficiently to risk huge investments and the reputation of their 5G services – that cloud-based offerings will deliver the same performance as integrated RANs in demanding environments (very high traffic and user numbers; very low latency, for instance). That delays the start of large-scale deployment, that in turn will drive scale into the ecosystem.
- Many operators are not deploying 5G just to increase data rates and quality of service for their existing services and users – they also want to expand their revenue streams with new enterprise, industrial and mission-critical services, which will make even heavier demands on the vRAN.
However, a group of advanced operators are starting to put timelines on macro vRAN adoption, spurred by the progress of greenfield providers such as Rakuten, and the innovations of a small group of early adopters such as NTT Docomo and SK Telecom. And the ecosystem is swinging into action too, to accelerate readiness of the cloud-native core and even the 5G vRAN.
The open RAN movement has proved very important for lowering the barriers to non-traditional vendors and integrators that are bringing new innovations and insights into the tight-knit RAN circle. In particular, it is providing a bridge for companies from the IT and enterprise industries, which have dipped their toes into MNO waters in the core, to expand into the RAN. In this week’s issue we examine how HPE, Dell and others are getting increasingly active in 5G, with their initial focus on the core – which is a more natural candidate for the cloud than the RAN, as it can be fully implemented in software. But they are also starting to formulate strategies for the RAN, as HPE announced last week. They will need radio partners of course, but they will have various points of entry to this large market – in the cloud infrastructure, especially where it relates to edge computing, which will be crucial to enable highly distributed 5G RAN and core; in supporting network-as-a-service platforms; in providing key technology components such as a RAN Intelligent Controller (RIC).
But if the MNOs have a learning curve to climb before they are fully structured and equipped to go cloud-native, so do the companies that pioneered virtualization and open platforms in the enterprise. Telcos procure, deploy and run systems very differently from enterprises or IT network providers, and their networks have different burdens to carry. VMware, Red Hat, HPE and others have gained significant presence in the fledgling 5G vCore and vRAN segments, as well as in other important domains such as white box-based transport networks. But have they tweaked their platforms for telcos or transformed them, and have they done enough that their platforms can support a vRAN that can be pushed to the very limit of 5G’s powers?
In the inevitable hiatus in which vendors and non-greenfield operators adjust to the new realities of 5G, there is a gap that may be closed by their efforts, or may be filled by alternative players. The deepening cooperation of Intel with Google and its Anthos managed cloud application platform is an example of the developments that could strike fear into the hearts both of established mobile network vendors and challengers from the data center industry. Microsoft’s acquisition of Affirmed and Metaswitch is another, and the Windows giant could be looking for an equivalent buy in the RAN area, to enable it to offer 5G-as-a-service – to enterprises first, no doubt, but potentially for macro MNOs in future. The ability of the webscalers to scale up their operations and to optimize their cloud platforms cannot be discounted in this race, and this may see mobile network vendors forming close, if sometimes uneasy, alliances with IT majors in order to accelerate and enhance their own cloud-based core and RAN progress.