Telcos will never achieve full network autonomy or business change by layering AI agents onto monolithic legacy systems, says TM Forum in the second half of its discussion with RCR. Telcos are not special cases, it says. They must rebuild their IT and network foundations around standardized, composable cloud-native architectures.
In sum – what to know:
Universal blueprints – TM Forum, compelled by China Mobile’s work with Huawei and ZTE, is developing vendor-neutral AN blueprints to standardize interfaces and workflows across telco environments.
Legacy IT revamps – Telcos must first replace rigid IT stacks with modular cloud-native systems first; TM Forum warns that AI layered onto legacy infra will only accelerate the same processes, and not drive change.
Same transformation – The industry is overly-complex, and a little self-obsessed, and should learn from other sectors to simplify operations and enable composable automation – and its stock-in-trade to shine.
Note this article is continued from here.
Where were we up to? Ah yes, China Mobile, and its drive to standardize developing autonomy frameworks, which are starting to combine a set of functional network “scenarios” into a broader domain exercise, across its Huawei and ZTE bases in Guangdong and Xinjiang – which has, to an extent, followed and informed the work by TM Forum to do the same. And George Glass, chief technology officer at TM Forum, has just suggested to RCR Wireless (see part one) that telcos should go back to basics, first, to rip-out their legacy IT stacks in favour of modular ‘cloud-native” systems – brand new tractors, or just faster horses, as Glass puts it. Do it properly, or what’s the point?
As well as stringing-together point solutions then, TM Forum is looking to abstract-away vendor idiosyncrasies – which advance network technology on one hand, but make it complex, often proprietary, on the other. “China Mobile is using Huawei in some places, ZTE in others.” The ‘technical solution package’, as set out by TM Forum (as discussed at DTW in Copenhagen next month), includes a description of the function / scenario, plus the architectural solution, workflows and interfaces, and a benchmark case with outcomes from a member-telco. Initially, the package was presented as a specific operator configuration of a specific vendor setup. The task now is to “normalize” these operator/vendor architectures into industry blueprints.
Glass explains: “As we normalize that, we will abstract and standardize those interfaces, irrespective of whether you’re trying to correct an IP router from Ericsson or Huawei; the instruction is the same. So it becomes vendor agnostic – just because nearly every operator dual-sources their network. They have to manage different interfaces, different commands, different providers – which is a complexity that is not required any longer. We eliminated that in the IT space with ODA when we transformed the IT architecture. We’re taking that thinking and applying that into the network space and transforming the network architecture.”
Regarding the China Mobile experience with Huawei and ZTE, he says Ericsson and Nokia are now building “similar patterns with non-Chinese operators – including with European operators, too”. Fault management is its first full AN blueprint, it seems – “wall-to-wall”, vendor agnostic. “Any operator with any vendor,” he says. “We remove the vendor interfaces; we abstract and normalize, and make them ODA compliant – so anybody can build it as a capability.”
Simpler foundations
The ODA reference here is to TM Forum’s base-level ‘open digital architecture’ model for telcos to replace rigid old OSS/BSS with flexible cloud-native IT (and subsequently network) architectures – and to build tractors.
He says: “You have to go from the base up to the top. You’ve got to build the foundations correctly. Data needs to be in the right format. It needs to be accessible; a uniform canonical data model across the organization. You don’t want to be doing data translations in and out of applications and components.
“You build on that, and then look at the functions or capabilities you need, and how to encapsulate and expose those – build once, reuse multiple times. And then: how do I deliver a customer experience that is not just about reinventing existing processes? You’ve got this autonomous capability, right – so how do you remove people from the process altogether?
He explains: “It’s not just about making processes go faster; it’s about redesigning how you manage the enterprise. You’ve got to get the foundations right, which is why we keep going back to ODA – the decoupling of the architecture, the encapsulation of business capabilities exposed by APIs. Which might be rest-based, event-based, agentic – but they’re standardized. And then you construct customer experience and business capabilities from that composable architecture, which drives reuse. And you show how to do it, based on what others have done – that’s where this hits the sweet spot, and why people are so excited about what TM Forum is doing.”
Is that a problem? Are telcos looking to deploy AN point solutions on top of monolithic legacy IT infrastructure, rather than going “wall-to-wall” on an ODA revamp, and reaching for proper business transformation? Are they saying, ‘George, this doesn’t work’, and you’re saying, ‘well, that’s because you’re not doing it right’? “Well, we’re trying to avoid them saying, ‘George, it doesn’t work’,” Glass responds. “We’re showing that it is proven, and how it will work. We were talking to a group CTO recently, and he thought, initially, he was transforming his network, and then realized he needed a composable IT architecture to manage and control it. That’s where we are trying to join the dots.”
Which is where the line about horses and tractors comes in. He adds: “If you leave your legacy and just start putting agents on top, you will not get to level four. You will automate your existing practices. You’ll get faster horses, but you won’t get tractors. A lot of our members are like, ‘I’ve got a little money, and so I’ll train my horses’.”
Inherent complexity
That ODA-style IT transformation sounds like what every enterprise sector is grappling with. Does the telco industry have a snowflake syndrome – like the problem is uniquely-telco, whereas it is a universal enterprise problem in the first instance, based in common-or-garden enterprise IT? And that the network complexity is unravelled on top?
“Exactly. That’s exactly what we’re saying. You have to step back from it. We’re trying to solve this problem for the telco industry, but we’re not doing it with telco blinkers on. We’re looking at what finance and manufacturing are doing, and learning from those AI patterns, and applying them to the telco space – and not thinking like telco is something special. A lot of it is very similar to any other industry – you create customers, add products, order products, send bills. Where there are faults, you repair the faults. Our product just happens to be a set of network connectivities, plus some devices. But a lot of the patterns are the same.”
The conversation gets a little philosophical, perhaps; the same question is asked another way. Is this industry a little self-regarding, wrapped up in its own radio genius – where it just needs to get real? “There’s an element of that. Many of us have been here all our lives, and don’t know anything else. You notice it when people come into the industry from the outside – that, yes, it’s complex, and, yes, it is critical national infrastructure; but we have overcomplicated it, and we can simplify it. We need to be very robust about standardization, and about retiring legacy products. Because if you get rid of those, then a lot of the complexity goes. People hang on to stuff for too long.”
Familiarity and sentiment gets in the way – and kills business, faster when tech change accelerates. Glass adds: “Organizations have the architecture and systems they deserve. If they tolerate legacy, they will have legacy. If they don’t, then they can move on. It is a mindset change.” But telecoms is necessarily complex, too – right? And there is inherent good in that. Simplistically, the point, we suppose, is that its back-end systems are broken, or not fit-for-purpose – to unleash a new wave of telco pyrotechnics. Is that right? “Yes, there is a complexity that gives us an edge – which makes networks work. And networks themselves are complex. But we over complicate.
“Things that could be standardized are not standardized. “It’s like having 20 different plugs around in your house – different for the refrigerator, telephone, alarm clock. Some are 240 volts, some are 120 volts, some are 55 volts; they have two pins, three pins, four pins. It’s like, you can do all of that, and justify why it is like that, but your house is full of sockets, and it is frustrating whenever you go to plug anything in, and unclear if anything new is going to work. That’s the situation we’ve got in the telecoms space.” What seems clear is China Mobile, plus certain others, are not snowflakes – whether because they know what’s at stake, or they’re just tired of searching for the right plugs.
