There is no single, reliable definition of what constitutes a ‘zero touch’ network. Most would probably agree that we’re talking about a network where high availability, reliability and services delivery are all automated, leveraging the principles of NFV and SDN. In other words it’s a network that monitors, manages and heals itself, and lets its customers pay for what they need with no human intervention, all through the power of software.
Not all are agreed that such a thing is even achievable, not at any rate in the near future. “I’m not sure there will ever be anything that is totally, 100% zero touch in terms of an end-to-end hybrid-cloud network,” says Atchison Frazer, worldwide head of marketing at SD-WAN vendor Versa Networks.
“There are way too many legacy devices and technical debt architectures out there, in addition to increasingly complex security requirements, for that to happen in the next five years.”
Opportunities exist to go some of the way, not least in the provisioning of wide area network services, believes Frazer. “You’ve got the zero-touch provisioning [ZTP] aspects of SD-WAN,” he points out. “Other than physically plugging in a WAN device, the turn up of features, functionality, policies, monitoring, reporting and analytics of every WAN device is zero touch, and that is applicable to 100 sites or 1,000 sites. Then once ZTP is established, ongoing policy updates and enhancements can also be zero touch, or at the very least minimal touch.”
Zero touch networks are still a relatively new concept, agrees Steve Foster, business solutions architect, UK and Ireland, at vendor Riverbed Technology.
“As a fully automated and orchestrated network, that modifies and adapts itself based on carefully constructed ‘intent’, they require complex capabilities and network architectures with cloud at the core,” he claims. “As such, the only companies currently capable of successfully utilising zero touch networks are those who have built their data centres from the ground up and have the need to accelerate network transformation to take advantage of new technologies such as 5G.”
Even the biggest network players face significant challenges here, says Foster. “While they can deploy zero touch networks today, they can only do so in silos, and upon interaction with any other providers’ equipment, the networks will falter and will have to revert back to traditional network systems,” he adds. “This is because successful zero touch networks will require an open standard between vendor infrastructures which still hasn’t been fully ratified. For zero touch networks to be widely adopted, we believe there needs to be a shift in the next generation of equipment in order for it to natively support new types of management system."
There are, notwithstanding the absence of definitive and universal standards, a number of carrier names in the wholesale space who have made progress towards a hands-free goal. “We’ve come quite far with zero touch,” says Mattias Fridstrom, vice president and chief evangelist with Telia Carrier. “But we’re not quite there yet. There’s still a certain amount of manual intervention to check it’s going to be right. We do let machines configure services, but just before we release them into the network we look to see everything is correct. What every carrier struggles with here are the inventory systems. When can you really trust machines to handle all that? It’s been really hard too to keep track of cross connects without manual intervention.”
François de Malartic, senior director of marketing data for the international carriers division at Orange, says his organisation has also been busy working towards zero touch, particularly on the all-important cross-connect side.
“We’re working with the world’s major carriers, through the MEF, to establish standards here,” he comments. “We’re directly working with others who are advanced in this area – people like Telefonica, Colt, AT&T. We’re doing PoCs with them.”
He says Orange has recently carried out a soft launch of zero touch services with 20 or so of its customers, called Internet Now.
“This makes connections on our network possible in two or three minutes – that’s real-time ordering,” he explains. “If this goes well, we’ll launch it to all our customers in May.”
Orange’s ultimate aim is to develop APIs on its ordering system for interconnection with all its main carrier partners, along with further APIs to interconnect with data centres. The goal is significant.
“Ultimately by making our entire offer available from a portal will enable to us to sell where we can’t reach today,” outlines de Malartic. “Zero touch is key to us in terms of future development. For us it’s a way to access a larger customer base. Today we’re mainly about Tier 1, but in the future we’ll reach out to Tier 2 and 3. We’ll be able to sell to small ISPs and operators, in places like Africa. Without this kind of thing, carriers will lose market share.”
It appears that most carriers initially approach the zero touch concept with the aim of internal efficiency, then branch out from there. Fahim Sabir, director of architecture and development for network on demand at Colt Technology Services, explains that was certainly Colt’s standpoint when it started on software-enabling its network at the beginning of the decade.
“We’re now onto out third generation of it with our IQ network,” he says. “We’re able to give zero touch power to the customer with the SDN investments we’ve been making for years. We’re also looking at AI-related initiatives as well, with network health and customer experience use cases. It’s another form of automation. We’re looking to get more out of an RPA model.”
So does AI-driven zero touch create a future where carriers employ robots for almost everything, disintermediating the entire human element?
“If you look way down the line, you can see the need for fewer people,” admits Sabir. “But I don’t see the human element entirely going away. You’ll always need people to install equipment and dig fibre. And even if customers can self-serve, a human side to it is sometimes important. If a computer says ‘no’, you might want to talk to a person about why. It means humans move up the value chain to focus around more valuable activities. Automation handles the grunt work. Machines can handle this because they are more predictable and consistent.”
With true zero touch operation probably still some time away, the pressure is on carriers to make ‘lighter touch’ network operations function, says Tom Footit, senior director of product management at Accedian, vendor of virtualised software tools for network-wide visibility.
“Challenges include being able to have continuous, ubiquitous and accurate information on how the network is operating, being able to analyse, in real time, the performance issues and determine a course of action, by leveraging AI and machine learning algorithms and being able in real time to make changes to the network based on the results of this analysis,” he says.
Troubleshooting is an important zero touch use case, agrees Carlijn Adema, head of marketing and comms at Nokia’s Global Services business unit which has worked on deploying ML and AI to solve network problems in a zero touch way. “We’ve been doing a lot with robotics and RPA,” she says. “When we move to automation, we can work five times faster and 100% correctly. That includes everything from alarm monitoring to restoration and everything in between.”
She says Nokia has moved on from simple rules-based RPA to AI that can learn from its mistakes.
“That’s a big difference,” she believes. “We’ve worked with a carrier in Brazil who wants to work on eliminating network issues before subscribers are impacted. ML detects anomalies, find the root cause and prioritise which actions to take. We can inject AI into the carrier network, based on what their concerns are. We’re also building a new digital architecture so that we’re ready for managing 5G. We’ll need to be more agile to manage the complexity of IoT.”
Fridstrom of Telia Carrier believes that zero touch networks are an ultimate inevitability in the coming years, but the industry might as well face the fact that no one is really there right now.
“Two years ago people were questioning if it was worth the investment, now it is clearly really, really worth it,” he concludes.