Colin Bannon, CTO of BT Business, talks to Annie Turner about the journey to commercialise NaaS, from redefining resilience and controlling latency through dual-homing, to data’s sovereignty in motion
Colin Bannon has been CTO of BT Business for eight years. He describes BT’s emerging global fabric in colourful terms, providing a clear picture of how cloud providers and innovative network operators complement each other. Or perhaps how telcos can augment cloud services is more accurate.
Bannon says, “This is as big as moving from analogue to digital…reflected in skills, culture, commercials, economic models, service and use cases for our customers to stay relevant for the future. Managing the transformation of the old to the new. And the last bit, and perhaps sometimes the least important, is the technology itself.”
Even so, “We’re building like fury right now,” he says. “Think of it on two layers. One is the physical infrastructure, the trucks and the equipment and the hardware. The racking and powering up and testing all the hardware. That’s going on geographically around the world right now. There will be more and more physical sites as the year goes on, in 2024 and 2025,” he states.
“We’re laying successive products on top of each node; additional software releases with new functionality and new products. We will do subsequent software releases…There is no end to this. We continue to innovate. There’ll be new features dropped in because now it’s like cloud.”
Could he have imagined this, say 10 years ago? “No. If you look at the innovation and the speed that the hyperscalers are doing on their platforms, I think it’s incumbent on the telcos to have equal ambition, and make sure that we are innovating equally to stay relevant to our customers’ needs,” he replies.
From coconut to avocado
A big change is that for years, data centres were the hub. Now they are a spoke. Formerly most corporate traffic was scrubbed at a data centre before it was allowed onto the internet. Bannon says, “You’d trombone all your traffic. People spent a lot of money building big racks of firewalls and load balancers where they’d have their apps etc.”
Now customers’ traffic needs a specific reason to go to data centres, otherwise it goes straight to cloud or SaaS. Bannon explains, “The concept of perimeters has changed. The network was like a coconut – hard on the outside, but all soft and watery inside. Once you got through that data centre and the firewalls, it was a trusted network internally.
“Now we’re dealing with networks that are more like an avocado, with less defined perimeters – people working from home, on their mobile, coming in through an ISP or whatever – but the applications themselves and the concept of zero trust have a hard core, like the stone in the avocado.”
The shift from ‘coconut’ to ‘avocado’ is a massive change which opens up new opportunities. Although we talk about ‘cloud’, networks rely on physical interconnection points – an IPX, carrier hotel or carrier neutral facility.
Bannon says, “There are business opportunities in being resilient and super-efficient.” He wryly points out that backhoes on plant machinery are one of telcos’ biggest problems, as someone is putting one through a fibre cable somewhere almost every moment of the day. BT designed its network as a mesh, so there are always multiple options at the optical layer if a link fails.
Somewhat ironically, automation can also destroy resilience. “Most of the big outages in the last year have been caused human error then automation magnified the blast radius, pushing something that was wrong to every box,” he says. In one instance, this included disabling the code reader on the door to a data centre that had to be accessed to fix the issue.
Bannon says, “That’s why an outage can take 12 hours now, because even if they know what the problem is, they don’t have that remediation.” BT’s way of avoiding such disasters is thinking “about the old school techniques of diversity, not just resilience. Failures in cloud availability regions are an excellent case in point.”
Dual-homing – more than twice as good
Say a company with a data centre in Japan send its workloads to, say, a carrier-neutral facility in Korea where a hyperscaler has an exposed network edge. In the event of failure, which could be for a multitude of reasons, common practice is to default to the next nearest route.
To access the application running in Korea, traffic is sent back to Japan via an undersea cable then onto Singapore via another such cable. Once the application has been accessed, the data needs to travel the reverse journey because, as Bannon points out, “Customers do not pay cloud providers to have an image of their application in every availability zone, they tend to just put one in a single region and think that’s resilient”.
The return journey could result in latency of between 200 to 250 milliseconds (ms) when already it’s likely that the customer’s experience has degraded, for example, with screens timing out.
Bannon states, “The way to fix it has to be additional diversity – and what we’re spending our CapEx on. Being dual homed, via different providers such as Equinix and Digital Realty, into that hyperscaler’s data centre in Korea. We’re doing this for Microsoft, AWS and Google, etcso rerouting is just a street across town with a 5ms failover, not going through the sea four times. That means you have resilience and performance.
“This greater robustness and differentiation brings real quality of experience that the cloud providers can’t fix themselves. We’re identifying needs in the market and solving for them as only service providers can. That’s just one example.”
Bannon says, “This is a differentiating factor within our network and part of our NaaS offer…Cloud providers have multiple options for resilience, shared or dedicated, and we’re making sure that we build and expose them digitally to our customers, making it really easy – a key click away – to order and change, to have visibility and control. That is significantly different to the way we do business today. And to be fair, most service providers have yet to go through that digital journey.”
Injecting intent for NaaS
That’s not all. Bannon continues, “We run our network on a modern segment around a core that has an abstracted path computation engine. The network is centrally controlled.” Previously routes were calculated, hop by hop with each hop across the core inside a ‘black box’ over which customers had no control. Nor had the black boxes any knowledge of the packets, they just moved them efficiently.
“Now, the central control…opens up fine-grained, micro segmentation and control,” Bannon says, giving BT total control and embedding business intent throughout the journey.
Why is this important? If a firm in Germany sends sovereign data to a data centre that data cannot legally leave that country. “But that backhoe goes through the fibre and the SD-WAN reroutes or reconverges, maybe peering through its partners via Russia then back into Germany. That’s not a great look today,” he says, as bad actors might well decrypt and steal or tamper with data.
SD-WANs are at the mercy of the ISP underneath and the internet was designed to withstand nuclear war. Hence “the paradox is that the internet is geographically ignorant. It is inherently viral in how it reroutes around a failure but we don’t have much determinism in how it reroutes,” Bannon observes.
Regulators are increasingly interested in what happens to data in motion, not only the sovereignty of data at rest. Bannon says, “We’re solving for being able to inject a business intent, such as to geofence this data within a country, as our service. That’s just one example of business intent.
“I don’t know a fraction of the business intents that will be thought up over the next 10 years. Having a platform that is programmable will make it relevant for future challenges. We need to manage the paradox of something that is inherently uncontrollable in a world of deglobalisation.”
Power to the port
BT Business is moving away from multiple racks with multiple stacks for each network to running multiple cores on a single stack. This helps maximise return on capital for shareholders and customers, Bannon explains. For example, a multi-service core and a multi-service edge is far more energy efficient, but also BT abstracts the concept of the port to the protocol and stack at the edge.
This allows customers to switch between services without the operational palaver of ceasing one service and provisioning another, involving engineers visiting sites, hardware changes, BSS and OSS. Instead, “The port retains the data, information and relationship with the customer to the lifetime of that contract. By abstracting the protocol and creating a software-defined edge, we can spin up whatever protocol they want to apply to that port or multiple ones,” Bannon says.
“All of a sudden, it’s a plugin that’s available to all the platforms. It cuts down on development, time, capex, code writing and testing. And when I talked about culture, this is genuinely a step change across the board.”