One advantage of a cloud-based network architecture for operators is the ability to update and enhance that network more quickly than what was possible with legacy platforms, a feat that Verizon is using to seamlessly launch new services, serve new market opportunities and continuously update its network.

Adam Koeppe, SVP of network and technology planning at Verizon, cited these efforts in a recent blog post that underscored the operator's growing use of cloud-native principles. This includes the operator’s initial foray with its own Virtual Cloud Platform (VCP) that runs on OpenStack and initially hosted and supported virtual network functions (VNFs). That later evolved into moving toward its newer Kubernetes-based OpenShift platform from Red Hat that supports container network functions (CNFs).

The VCP continues to be the anchor for many of Verizon’s core network functions, but the latter is increasingly the platform behind new services. Koeppe, in an interview with SDxCentral, explained that this cloud-native architecture has changed how networks are managed.

Minimizing downtime while maximizing scalability

“When we look at the architecture and then the solutions that we use to run it – like a Kubernetes – we are trying to find things that can minimize the downtime when there are either failures, faults, improvements, upgrades, software releases — whatever the case might be,” Koeppe said. “Having a web-scale environment and being able to push software improvements rapidly to elements on the network with very little downtime is one of the key benefits of this entire architecture.”

Koeppe said this was superior to the old model that required more time and introduced more complications.

“In the old days you’d be physically going out to places and doing a very antiquated process of upgrading software over the network. This is a much more rapid environment,” Koeppe said.

That increased agility also supports the scaling of new services as they flow into the market. Koeppe pointed to Verizon’s rapidly growing fixed-wireless access (FWA) service as an example of that scalability.

“If you think of customer-facing applications — so a set of functions that I'm running in my cloud to provide fixed-wireless access services that we sell to customers — they probably have a different growth rate than some of the transport functions that I have in a virtualized environment. I would manage [those] very differently than something that's facing the customer,” Koeppe said.

This flexibility also extends to the carrier’s deepening 5G radio access network (RAN) architecture. Koeppe explained that Verizon’s more than 12,000 virtualized RAN (vRAN) sites are “basically a Kubernetes cluster.”

“We've got basically agnostic hardware in the cell site that has that virtual [distributed unit] running on it and that looks like one of our cloud clusters,” Koeppe said. “And there's thousands of them.”

Operators want to maintain cloud control

Koeppe also reiterated that while a cloud-native architecture allows for more dynamic network control, Verizon plans to maintain control over those network assets.

“Our goal from the start and still today has been we're gonna have a network cloud that we run our core functions on; this is the network that faces the customers,” Koeppe said. “We are going to be responsible for that from beginning to end, the whole package. So placing those network functions on somebody else's infrastructure, or cloud, is not even part of the equation.”

This mantra was echoed by rival AT&T executive Igal Elbaz, who told SDxCentral in a previous interview that “AT&T’s 5G mobility core runs on AT&T hardware in the AT&T data centers and operated by AT&T.”

However, both executives did note there was a place for public cloud to also be mixed into specific environments.

“We’ll probably see scenarios where a multicloud environment comes into the picture and pretty much almost always tied to an enterprise need,” Koeppe said. “If you've got an enterprise customer that is trying to manage some of their applications or workloads on a private network, on a public network, whatever the case might be, we can facilitate that type of architecture as well.”

AT&T’s Elbaz said these sort of architectures also allow operators to better take advantage of tools and automation developed by hyperscalers.

“At the heart of this is how can we take advantage of the hyperscaler investment in tools, automation and cloud capabilities, because that’s their business and they are really good at it,” Elbaz said. “And knowing that this is how we are going to run the network going forward so why not take advantage of this.”

Bringing hyperscalers into the mix

That investment is growing as hyperscalers look to gain more business in hosting telecom networks. A recent report from Dell’Oro Group forecast that hyperscalers like Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP) would be hosting 9% of 5G standalone (SA) workloads in their public cloud environments by 2027.

Ishwar Parulkar, chief technologist for telecom and edge cloud at AWS, explained to SDxCentral that the hyperscaler is well aware of the “carrier-grade resiliency” needs of operators.

“The cloud was designed to be highly available,” Parulkar said. “The enterprise workloads might be one category, but the cloud runs government workloads, we run financial workloads — the Nasdaq Stock Exchange now runs on AWS — so we have a lot of other very critical workloads that have been running for a while. … And the requirements of those are a very high bar. So we continuously work on making the cloud very highly available and building more and more capability into it to make it highly available.”

Parulkar said that the challenge for hyperscalers is to now change the mindset of operators.

“From hardware rack-level redundancy to looking at availability zones and having availability and resiliency at the software layer is what has changed," he said.

"And sometimes it takes a little while to accept and understand the change; understand you need to look at the same metrics or five-nines in a different way,” Parulkar said. “But these are some of the things that we are actively working on. And once we go through a detailed analysis, the operators see that.

"At a basic level it might not be so evident how the model is different, but there is a difference in the model from going from hardware redundancy to software-layer reliability.”

That detailed analysis will also need to have a financial component. Koeppe stated that Verizon’s virtualization efforts have already proven fiscally beneficial for the operator, and that making larger network changes are typically aligned with broader network evolutions.

The economics of digital evolution

“The cost savings for this and the economics behind having the virtual cloud platform prove themselves out very, very quickly. And that's especially true when you have a catalyst for deployment,” Koeppe said.

“We talk a lot about our vRAN; we've talked a lot about moving to a cloud-based platform in the core. When you have a kind of a large technical shift, or you've got a series of equipment that's approaching end of life, or you've got a partner or supplier that is challenged and you're going to replace, when you have that catalyst where you can move to a cloud platform, core or edge or far edge, it's very clear to do the financial assessment of the value this type of architecture creates.

"Because, he explained, "you have today's world and what you're going to go to, and you'll know what your scale is, what your quantities are, what your sizing is to do a really effective financial assessment. We've seen this architecture exceed our expectations on the overall economic picture for value. So that's been a really good story.”