BT and Colt spell out critical need for network platforms

Telcos that have yet to build software-defined, programmable network service platforms need to do so urgently. That was the message from Colin Bannon, Chief Technology Officer, BT Business, and Fahim Sabir Director, Digital Solutions, Colt Technology Services, speaking on a panel during Telecom TV's ‘Telco as a platform’ online summit.

“I don't think that we will have a role to play unless we ... start exposing what we do as digital platforms,” warned Sabir. “If it doesn't start now, then … you're also going to be trying to learn all of the lessons about trying to run a platform while competing in a market where there are lots of platform-based operators.”

And a fully programmable network platform will be essential if telcos are to make the most of AI, including agentic AI.

“When AI starts to hit the network … you really need programmable everything in your network,” said Bannon. “Everybody talks about AI, but they don't talk about the network implications of it. If you haven't started building your platform, you're probably going to miss out.

That's because adopting a platform model is a long and difficult undertaking for a brownfield telco, according to Sabir and Bannon.

Going against the grain

“Moving towards a platform type environment is an incredibly uncomfortable thing for a traditional telco to do. It goes against basically every measure of success that we've had in the past,” said Sabir, who added there is no room for half measures.

“It's one of those things that you go all in, or you don't do at all," he added. "The biggest risk, is that we go halfway, so you don't get any of the benefits of the new world, and you lose all of the benefits of the old world.”

Indeed telcos that haven’t started the process, “may only have the option to milk their legacy platforms and manage them for cash rather than manage them for innovation,” according to Bannon. “Because I just don't think putting a thin layer of digitalization on top of a really old tech … stack will deliver the platformatization difference that our customers are going to need.”

A successful platform allows internal teams, customers, developers, enterprises and other carriers to unlock network assets and spin up new services.

Centralizing control

But it requires profound technological, cultural and operational change, with a centralized architecture and a strong operating model usurping traditional product silos.

“Platforms are only useful when they can be controlled,” stated Sabir.

“It's got to be horizontal, and the data needs to drive this,” according to Bannon. “This is not a democracy. It will require centralization, where the architecture team are more empowered [to provide a] common architecture and a common platform, rather than the individual development teams that were tied into the products previously."

When creating a network platform, for example, BT “ripped up the blueprint and the stack and built from scratch a converged, microservices-based, modularized, componentized platform where there are reusable components, like security, single sign on,” etc., explained Bannon.

Instead of building services within the parameters of a specific use case such as an end-to-end Ethernet service, for example, telcos can add and combine components on the platform to flexibly create services they may not have foreseen..

In addition, teams no longer need to test launches against E-LAN, E-LINE, MPLS, Internet and mobile, services, explained Bannon. “Now we test it once against the platform. You start to abstract concepts of the platform, so the port and the protocol - the underlay of our network - is now programmable, and we can spin up new services just like a platform.”

In the beginning was the API

With a platform model APIs are the starting point of product creation rather than an afterthought. They become “part of the non-negotiable stack that makes up what telcos have to offer,” according to Sabir.

This represents a profound change in how product teams operate.

“When you move to platform-based thinking … it becomes an intensely uncomfortable conversation for a product manager, where you used to own a full stack and the box that was attached to it," according to Bannon.

Data also needs to support the use of APIs.

"A lot of these APIs have to be data-driven in terms of what value they provide to the upper layers of the stack. And if you don't have the right amount of right kind of data … then you're going to end up with silos, and the APIs will not give you that kind of value."

Managing customers

Internal operational efficiencies and costs savings are important benefits of an automated, component-based platform approach to building services. It can also bring tangible benefits to customers.

Sabir stated that it offers “granular level of control and programmability that’s absolutely critical for a certain segment of our customer base."

Not every customer will be aware of an operator's switch to a platform model, however. Some customers simply want connectivity rather than the ability to compose new services and that is all they will continue to buy.

Where telcos need to be careful is with large enterprise customers for managed services, which are seeking outcomes to business problems.

Here, what Bannon described as “a more lights out or automated model” could impact managed services for large enterprises that typically rely heavily on people for sales, support and delivery.

“We feel a great responsibility to not impact the operational resilience of our customers," he said. "It should be greater than the sum of the individual constituent parts, rather than taking away from the managed services.” This too, he acknowledged, “requires much innovation and thinking.”