On a recent episode of Zero-Touch Live, Boost Mobile's Dawood Shahdad, SVP of Hybrid MNO Network, and Sruthi Nair, Director of Voice Core Engineering, spoke with Rakuten Symphony's Anshul Bhatt about the journey of building one of the few operator networks running fully cloud-native on public cloud. In this week’s Zero-Touch newsletter, we dive deeper into the role cloud-native plays in powering autonomy and the essential building blocks that must accompany it.
At DTW this week, it seemed every stakeholder was eager to discuss the AI and autonomous network pursuits that would let their networks adapt, scale and support growth.
Getting to Level 4 autonomy and beyond is about readiness and ability to execute on the impactful use cases that move the needle on business value. It comes down to what an operator can actually do in practice.
The announcements we're starting to see industry-wide can sometimes appear as overnight successes. In reality, they are years in the making. The strategies and building blocks in place to make these deployments possible are a necessity for fueling more ambitious efforts. Otherwise, telcos are stuck doing bits and pieces of AI but fall short of end-to-end transformations that power the real magic.
This is especially true of hybrid patchwork attempts where some functions are on physical network functions and hardware, and others are on virtual machines.
Boost Mobile recognized this years ago. As it designed its network, it had a simple but important rule: every network function had to be cloud-native on public cloud with no exceptions. It was a decision everything that would come after needed to be able to build upon.
What is truly-cloud native?
Being “truly cloud-native” is not the same as having AWS as an infrastructure partner but running apps on VMs. The network has to be able to drive itself and react in real-time. It scales up when traffic spikes, recovers on its own from outages and absorbs anomalies without breaking a sweat.
This real-time nature is only possible when functions can spin up in seconds. VMs that take a minute to boot can’t keep up with the needs of real-time autonomy. That’s why VM-based or hybrid-patchwork networks are non-starters. They limit how far operators can actually take autonomy. And no amount of AI deployed over top can change the underlying realities.
Traditional telco trappings do introduce some constraints here. Netflix doesn’t have to worry about voice, emergency calls, critical traffic and 3GPP rules. So web-scale assumptions like “just send the traffic and we’ll scale” can’t work in telecom network scenarios where video buffering is fine but dropped calls are not.
Still, operators will need to get out of their comfort zones. Boost Mobile rolled out the world’s first end-to-end 5G standalone core running on public cloud but getting there meant pushing vendor cloud-native systems past their limits, faster than planned.
Seeing the whole network
Boost Mobile’s cloud-native foundation simplified observability for logs, alarms, metrics., etc. network-wide. No matter what part of the network, data is collected the same way.
This is a contrast to traditional multi-vendor deployments where each vendor might bring its own EMS, requiring engineers to learn disparate tooling, monitor separately and manually connect the dots across systems to pinpoint issues.
But by rolling out one centralized, vendor-agnostic observability platform with a single pane of glass across the packet core, IMS and messaging, Boost Mobile is freed from the trap of vendor silos. Plus, tweaking the network or deploying a new CNF doesn’t require a netops retrain and additional engineering on new elements. Instead, they can use the time and resources saved to focus on new features and innovation.
This approach creates a new dynamic. Vendors supply the products and Boost Mobile calls the network topology and architecture design shots. Taking this extra responsibility onto its shoulders demands it has eyes over the whole architecture.
Today, it is one of the critical building blocks that turns cloud-native from just an architecture to something that can power business outcomes. It’s also one of the primary prerequisites for automation.
Advancing closed-loop automation
An isolated alarm isn’t the same as a real network problem. A single node may see IP fluctuation but that doesn’t necessarily mean there will be customer impact.
The framework that sits over the observability platform and reads each alarm in network context needs to understand this. It can leave low-impact issues for engineers to address eventually. It can also be programmed to act in defined cases.
For instance, if X alarm is triggered at Y stage, go to Z node, check for conditions and take action as required.
In practice, that may look like a UPF with an MG card failing and the framework taking the operational step to pull the noted out of rotation with zero human intervention.
Boost Mobile saw this functionality as a crucial building block. The old pattern meant its nimble team was logging in, spending 20-30 minutes on access and authentication, then running a 15-manual diagnostic process before experts even weighed in. By that time, the impact could be cascading. Automating periodic data collection and integration, and anomaly detection is what made it possible for network to surface these issues proactively.
Ultimately, Boost Mobile wants to be able to “talk to the network” to ask how its doing and receive intelligent reporting in response. Observability plus cloud-native plus modern monitoring are building blocks that will make it possible.
It’s what careful, deliberate steps toward a fully autonomous network North Star look like. Today, many use cases sit firmly in the Level 3 to 4 paradigm. The caution is warranted given the potential blast radius of automated action gone wrong in the high stakes core network.
Acknowledging the hard human transformation element
Technical building blocks are an outsized component of the automation puzzle but the hard work that goes into transforming human mindsets can’t be overlooked.
It actually must precede technical deployment.
It’s easy to forget there’s a customer at the end of every technology decision. Minimizing downtime and connection interruptions to maximize the customer experience become the unifying purpose. When all teams orient to these goals, the usual silos start to dissolve.
This can be easier said than done and a challenge area for every telco. Siloed teams may have a common goal in mind but end up prioritizing their own challenges or forgo collaboration in the name of a quick, albeit narrow) resolution.
That’s where the onus falls on leadership to align culture and technology, constantly looking three to five years ahead at the technical debt every decision makes. There has to be a willingness to try, fail fast and adapt.
It’s a necessary and worthy goal within reach for any operator with the willingness and mindset. It outweighs budgets and systems every day.
The payoff
Despite its criticality, cloud-native isn’t the finish line. It’s what positions operators to solve future problem sets, at global scale. For Boost Mobile, its network in the public cloud makes it possible to acquire or build data centers anywhere. Eventually, that could mean direct-to-device and direct-to cell calling with a seamless experience between terrestrial and non-terrestrial networks.
Of course, the public cloud element may not make sense for all. Readiness is the universal requirement.
In Boost Mobile's eyes, every operator can achieve this state and add to the collaborative collective that advances telcos industry positioning, even as new entrants bang on the market door.
So where to start? That will be dictated by an operator’s own priorities and willingness to get uncomfortable .
What conversations did you have at DTW that are helping to frame the building blocks your business will require?
.svg)
