Zero-touch networks are possible today, I’ve seen them – Juniper

Zero-touch has been lauded as nirvana for the telcos, but some might longingly look at the promise as unrealistic. That shouldn’t be the case according to Juniper CTO Bikash Koley.

It might be incredibly complicated, it might be expensive to start and it might be impossible for some legacy areas, but Koley point towards his days at Google where software ruled the roost. Zero touch networks were in place there, the dream is real. Take that sceptics.

“I know it is feasible because I’ve seen it,” said Koley, who’s previous job was Head of Network Architecture, Engineering & Planning at Google.

Of course it was an easier job for the hyperscale players such as Google. They didn’t (or don’t) have to deal with legacy technologies and have a different type of workforce. The operators have to deal with old-school hardware and a different set of skillsets meaning that while they can follow the lead of the hyperscale players, there has to be a very different approach.

Automation is never going to be a one-step win, but there are areas which can be automated first. The data centre is a perfect example as it is segmented. Koley highlighted an opportunity to play around with technologies in a specific area of the asset, nailing the process before moving elsewhere. It is keeping an eye on the long-term goal and having to accept that perfection is never going to be a possibility.

A good example is with the legacy technologies which are in place. Yes, it would be nice to automate processes here, but is it worth it? What is the cost benefit of automation comparing to the time period before retiring the technology? Operators may have to wait until technologies are being retired to live the automated dream as it is simply more cost efficient to put up, shut up and pay up for new down the road.

Automation is coming. Look at the hyperscalers, they did it because of the network reliability benefits and a more robust, reliable network is something every operator would want. Most mistakes on the network are human error, so it might just have to be a case of getting rid of this risk. Conceding this round to the machines might have to be the case.