How elegant software can make 5G networks more resilient

Telecom networks are part of the core fabric of our society: Any interruption can cause huge disruption. With the COVID-19 pandemic pushing more and more activity online, now more than ever, networks have to be available around the clock — a fact that is top of mind for mobile network operators around the globe as we move into the 5G era. 

Ensuring 5G mobile networks will meet that crucial requirement was one of the main themes of an online discussion hosted by Light Reading earlier this month on how to implement, deploy and operate a cloud-native core network. The brains of a mobile network, a 5G core, will need to support important new functionality, such as network slicing (dedicated connectivity for specific customers) and ultra-low latency (highly responsive) services, but without compromising on reliability.

“The key thing that I come back to and I hear most often and most consistently from operators is: It’s all about availability,” Gabriel Brown, principal analyst at Heavy Reading, said in his opening remarks. “It’s one thing if you lose Facebook or if you lose Zoom or any other application you use, but if you lose mobile core [connectivity], you…lose everything. And so operators are obviously intensely aware of that. They’re also rightly cautious about this big transition to cloud.”

As a pioneer of cloud-native mobile networks, Rakuten Mobile has firsthand experience ensuring that its innovative software-driven architecture will be at least as reliable as its predecessors. “As you get to learn and discover the benefits and advantage of cloud, you also get to discover the elegance of software and how, done right, software resiliency, software reliability can enable a network to achieve very high levels of availability,” said Tareq Amin, CTO of Rakuten Mobile, during a presentation in the online event.

Rakuten’s new network in Japan employs a microservices architecture that could theoretically support “an infinite number” of user plane and control plane functions, Amin added. By completely separating the user plane from the control plane, Rakuten’s 4G packet core has the flexibility to direct traffic to different data centers in different locations, as required. That means the network’s data centers can seamlessly support each other, handing over tasks to smoothly manage spikes and surges in traffic.

How Rakuten Mobile's elastic cloud network architecture can enable a network to achieve very high levels of availability.
How Rakuten Mobile’s elastic cloud network architecture can enable a network to achieve very high levels of availability.

Remarkable levels of resilience

Amin recalled how his engineering team initially experienced “many stumbles and many issues” with its pioneering core network, but “what I got very excited about is, within seconds, having the ability to move [a workload or a session] from one location to the other instantaneously and without user intervention.” This capability pointed to the opportunities for a pure cloud native infrastructure to achieve “remarkable” levels of resiliency, he added.

Rakuten Mobile’s network architecture is designed to enable its customers to enjoy unlimited data traffic.  But clearly the network needs to be able to cope with the demand that will result from such an attractive proposition. Amin told the webinar audience that the network, which launched in early April, is “scaling up very nicely” as Rakuten Mobile’s customers generate twice as much traffic per head as those using alternative networks.

The packet core is one of three layers in Rakuten Mobile’s network architecture, along with its 58 regional data centers and 330 “far edge” sites, which are about 30 kilometers away from the base stations (see diagram above). Later in the discussion, Sagiv Draznin, vice president of service experience at Rakuten Mobile, explained how these three layers work together, providing redundant storage and processing capacity as required. “It requires an almost …new way of thinking that you can actually have a workload go down and that’s fine,” he said. “In the era of cloud computing, hardware goes down: It’s fine… but the ability to recover and still have session continuation is very, very critical and the architectural three layers allows us to achieve that.”

Coming next – no more capacity concerns

Next year, Rakuten Mobile plans to deploy a standalone 5G core network developed together with NEC Corporation. Amin said that this core’s advanced microservices architecture, in which software can be completely self-contained, will give Rakuten Mobile automated elasticity. Once the new core is implemented, Rakuten won’t have “to worry about capacity anymore,” Amin explained. “That’s a thing of the past as the containers that you have manage themselves and … elastically grows based on demand.”

Although the new standalone 5G core will run in parallel with Rakuten’s existing core network for a period of time, Amin said the plan is for the new core to ultimately take over 4G functions. “We will take the 5G core, once built with all containerized functions and components, (we) will collapse all of the 4G functions, and in probably 14 to 16 months we are anticipating we will have one single converged core 100% pure cloud-native function on Rakuten Communications Platform,” he explained.

Other mobile operators are also working towards deploying standalone 5G networks, but the pace of adoption is set to vary considerably given the technical challenges involved. A survey of 150 operators conducted by Heavy Reading found that 16% of respondents plan to deploy this architecture in 2020, followed by a further 40% of respondents by the end of 2021. Noting that operators really need to have extensive 5G radio coverage to justify a standalone 5G core, Gabriel Brown described 2021 as the “takeoff year”, but cautioned: “I do think broad … global adoption may still be a few years out.”

Tags
Show More
Back to top button