Many mobile operators are now in the process of migrating their infrastructures towards more open platforms in the backhaul, core and even RAN domains using various technologies such as disaggregation, network functions virtualization (NFV) and software-defined networking (SDN). And some are just starting to move on from NFV to the more modern and agile cloud-native platforms, which implement network functions as containerized microservices rather than virtual machines.
The migrations are being driven by competitive forces, including hoped for capex cost savings, flexibility, freedom of vendor choice, future-proofing and speed of deployment.
In theory, 5G should allow a clean break from legacy but in practice there will be coexistence at various levels for some time, especially in the RAN and at the edge. There will be a lot of 5G Non-Standalone (NSA) deployments, deployed as an interim step towards 5G Standalone, as well as 4G and other legacy installations, plus intermixing of technologies in the backhaul and core.
It is at the edge or cell site where the inertia of legacy will be felt most strongly and yet where the potential benefits of disaggregation are greatest. There has been a concentration of power there among just three or four key vendors, often coming down to two, Ericsson and Nokia, in regions where Huawei has been banned.
While even some of the largest operators have taken the path of least resistance by continuing with the big infrastructure suppliers, a growing number, such as Telefónica and Deutsche Telekom, are pursuing open alternatives around Open RAN that promise to deliver separation between hardware and software across the control and data planes.
This is a step change that is taking place, as the big vendors support control plane separation but continue to peddle proprietary systems at the data plane level, where the data forwarding logic is still encrusted in ASICs. That is the state of play for many operators for which the shift to SDN has occurred at the control plane level, but without the implementation of NFV or containers to separate underlying data switching and forwarding functions at the data plane level.
This has already yielded benefits for many with the ability to set policies across a whole network via graphical user interfaces, rather than having to rely on lower level commands. This has improved productivity within IT departments, and for some operators Open RAN may extend these benefits out to the edge. They can then cherry-pick to some extent, selecting what they consider the best application software and components for each node, which they cannot do when the same vendor’s kit is deployed end-to-end.
It is true that the major vendors support Open RAN in theory, or even in principle, but operators are still locked in to their ASIC-based hardware at the data plane level, so that there is no move to cloud-native. This may simplify management but can also preclude operators from deploying more advanced features offered by alternative vendors seeking to compete on this basis. That will become increasingly important as 5G becomes more ubiquitous and operators compete on features and capabilities, particularly for those lucrative enterprise customers.
This leads to another point, the need to upgrade skills in line with the transformation to software-based disaggregated networks, which will require additional or different capabilities. In practice operators will have a blend of legacy and more software-centric infrastructure for some time to come, so will need to maintain a blend of skills to manage both, possibly training existing staff to embrace both.
The main point is to avoid skills, or lack of them, impeding migration to SDN and associated disaggregation across the whole infrastructure, from RAN back to core, now that operators agree this is the way to go. The problem lies not in the technical capabilities themselves, given that experience with virtualization and SDN already exists at the enterprise IT level. Indeed, many operators and large enterprises will already possess these skills inhouse and they can be repurposed.
It is at the higher level of integration and planning where new skills and perspectives may be required. Until recently most operators have often been able to rely on one or two monolithic vendors to handle integration and workflow management and indeed it is this comfort blanket that some are reluctant to give up for the seemingly unproven benefits of disaggregation and software-based networking.
For operators themselves, this skills conundrum is both a challenge and an opportunity. The challenge is to get their own house in order, while the opportunity lies in then offering a full networking service including integration to enterprise customers.
For the big vendors there is also an opportunity embracing virtualization and Open RAN and opening up to multivendor components across the data plane, becoming themselves masters of multivendor networks rather than just proprietary systems integration.
There will also be opportunities for existing and emerging specialists in systems integration, whose role in planning and deployment will expand considerably in the 5G era. It is true that standards bodies, such as the 3GPP, O-RAN Alliance and the Small Cell Forum (SCF), can smooth the path to integration by defining parameters for configuration, management and implementation of service level agreements, but the greater freedom of choice will still bring higher level challenges in selection, deployment and management of appropriate components and tools.
Some larger operators such as Deutsche Telekom have already embraced these challenges and developed the required skills largely internally, but many operators will require more outside help. The challenges will only intensify as 5G advances and network slicing becomes more widely deployed, requiring a balance between requirements of different applications and customers sharing the same broader infrastructure.