The benefits of cloud-native RAN include efficient scaling, dynamic orchestration of microservices and enhanced security
Operators are steadily working to become cloud-native from the core to the radio access network (RAN) all the way out to edge computing nodes. The rationale is two-fold: a cloud-native 5G Standalone architecture is seen as key to monetizing new classes of advanced 5G services while also helping increase efficiency (reduce opex) through automation and optimization of compute, network and spectrum resources. Open RAN—a standardized approach to RAN hardware/software disaggregation—has emerged as a vector for fostering network vendor competition and underpinning operator’s ability to innovate in the larger push to cloud-native.
In the recent Telco Cloud & Edge Forum (available on-demand here), experts discussed how the market is shaping up—is the high-level move to the cloud driving Open RAN interest and adoption, or is the ongoing standardization and deployment of Open RAN helping operators make progress toward longer-term cloud ambitions?
Before tackling the chicken-and-egg question, panelists first level-set on what cloud-native actually means for mobile network operators. AT&T’s Amy Zwarico, director of cybersecurity, put it this way: “It really means that the hard part of running an application in production like auto scaling, host hardening, they’re all handled by the cloud and it lets me as an operator reduce the cost of running my applications because so much more of that really difficult nitty-gritty work has been automated for me in a standardized way.”
CableLabs Director of Mobile Networks Mark Poletti carried on the centrality of the efficiencies that come up with running network functions in a cloud environment. “Cloud-native is sort of an environment that separates the hardware and the software and the compute and the storage and that disaggregates the virtual functions so that can be scaled separately to meet service needs and deployments and just time management inefficiencies.”
Moving away from bringing cloud efficiencies to the world of mobile, Viavi Solutions Regional CTO Chris Murphy expand on the innovative potential enabled by this paradigm shift wherein stateless microservices are dynamically orchestrated. “The gold standard is that each of these components are upgradable independently from each other. And the key to this is continuous integration, continuous test, and continuous delivery…I think these points will have implications for cloud-native RAN.” Murphy also noted the innovative potential associated with the Open RAN Intelligent Controller (RIC), essentially an application compute platform that supports the two-fold Open RAN thesis of new services and reduced opex.
“In the RAN intelligent controller, this is where having separation of concerns really comes into itself because there’ll be foundational capabilities in the RAN intelligent controller such as things like data aggregation, anomaly detection, things like location estimation. These will take place in the RIC, and those foundational capabilities will then drive multiple, in many cases, use cases to actually deliver performance improvements. So that’s where a lot of efficiencies will come, I think.”
Cloud-native Open RAN considerations—CI/CD, container management and standardization
With obvious benefits, both internal- and customer-facing, brining a cloud-native approach to a disaggregated RAN (or is it the other way around) does seem to be a priority for operators. While there are quarterly ups-and-downs in the market forecasts provided by firms like Dell’Oro, Open RAN is progressing, and numerous large, multinational operators have well-articulated deployment plans. Outside of brownfield operators, prominent greenfield builds—1&1 in Germany, Dish Wireless in the U.S. and Rakuten Mobile in Japan are the current marquee examples—are following this new architectural and operating model. So how do these two distinct but intertwined trends relate to one another; “Can cloud-native RAN accelerate the shift towards Open RAN?” Pongratz asked.
“I see it the opposite way,” Zwarico said. “I actually think that…once Open RAN is able to adopt the specs that are being developed by the O-RAN Alliance, I think that’s actually what’s going to drive the adoption of the cloud-native RAN…Until you’re standardized from an operator perspective, I’m not sure that I can achieve the full value of just deploying…the proprietary components and interfaces onto cloud hardware. I don’t think that gives you the full value.”
Poletti called one piece of the Open RAN vision and plug-and-play interoperability between hardware and software vendors, something that isn’t quite a reality today but, again, is part of the long-term vision. In that context, “you’ve got the vertical integration and the horizontal integration,” he said. “Vertical integration is the hardware, bare metal. And the you’ve got your virtualized functions and then you’ve got your applications. And that kind of has to be solidified first before you can do the horizontal integration of the getting the interfaces to work.”
Going past the integration to the actual management of microservices and cloud infrastructure, Poletti mentioned the important selection of a Kubernetes container management platform which are provided by the likes of Red Hat, Wind River and VMware. “If you’re an operator, having a common platform that can go through its lifecycle management at the same time your containerized network functions are having their product lifecycle. And if you want to mix-and-match and go to a different cloud platform, how does that work?…It’s an implication that needs to be ironed out, and I think that’s kind of working its way out in the industry.”
Murphy tied the whole question together around the hinge of the actual process involved in making best use of a cloud-native RAN and the potential for new service delivery and opex reduction that entails. Well-established in the world of enterprise IT and making its way into telecoms, a continuous integration, continuous delivery (CI/CD) approach allows for dynamic software management and, on paper, makes for a more responsive network.
“Someone’s going to have to do this,” Murphy said. “That’s a discipline which we get from cloud-native. And it also gives us an opportunity to find the issues that are becoming a problem before they become a big problem. That’s the real benefit here. And this becomes so important because we’re asking a lot from our 5G networks. IF we can’t deliver exciting services and exacting quality of service SLAs, then we won’t be getting the most out of the 5G project.”
When, where and how?
With benefits and implications established, when, in what type of deployment scenario and how do operators make this transition to a cloud-native RAN that follows standards developed within the O-RAN Alliance? Short answer from the panel, there’s still more standardization work to be done, operators need to come to grasp with not just CI/CD but also continuous testing, and the business case has to be airtight. As to where, Poletti suggested potential overlay or underlay of existing networks, or even for private networks. Murphy hatched another chicken-and-egg problem by suggesting the where will be determined by the necessity of cloud-native RAN to deliver X service or services.
Zwarico on the how painted a candid picture of how operators have to think about any sort of scaled RAN deployment, cloud-native, open or other. “The RAN is highly distributed,” she said, unlike core networks. Touching a radio site requires a truck roll, hardened computing infrastructure, fiber, etc…Unlike a datacenter, “The RAN is not a controlled environment and I think that that’s where there are logistic problems that have to be solved. Those are probably the bigger problems than how do I disaggregate it, how do I turn it into microservices? I think that part is the easy part. It’s logistics.”
