OTTAWA -- The telecom industry is moving toward autonomous networks, but Ciena says it is taking a different approach. As a supplier of optical, packet and software automation technologies, Ciena is eager to promote the "Adaptive Network" as a means to enable the network to react to unanticipated changes in traffic, connectivity and services.
In a recent interview with Light Reading, Brian Lavallée, senior director of Solutions Marketing for Ciena Corp. (NYSE: CIEN), says the Adaptive Network could assist service providers in managing and re-allocating bandwidth for the "glut of traffic demand" in upcoming 5G mobile networks. Also, he says this approach can help operators turn up services faster and "get rid of those mundane tasks people do over and over again that are error-prone."
At Ciena's annual Vectors Event in Ottawa, Ontario, Light Reading sat down with Ciena CTO Steve Alexander to discuss why some customers are leery of fully autonomous networks, and how Ciena is collaborating with service provider customers to prepare for 5G deployments with platforms like the Adaptive Network.
Kelsey Ziser: How is Ciena working with service providers to help prepare for the infrastructure, capacity and traffic management requirements for 5G?
Steve Alexander: Where we usually end up talking to customers about 5G is in three different areas.
One piece of 5G is the user experience, devices like your cell phone gets faster. The second piece that frequently is talked about is point-to-point microwave or millimeter wave, but it's the wireless edge/wireless access. The third piece that's discussed is 5G as the infrastructure for the Internet of Things.
All of those require densification at the edge of network. So more antenna locations, more fiber to more antenna locations, higher rates per site -- densifying the edge of the network and adding more and more capacity. So customers want to know how to put fiber closer to the edge of network -- the MSOs in particular have a project called Fiber deep which is replacing what is effectively the old analog coax plant with a fiber plant as close to the edge customers as they can make it go practically. If you already have fiber connections to cell towers, you're talking about how to move them from 1G connections to 10G connections – we even see some folks thinking about moving to 100G connections to handle the sheer volume of traffic coming to the network from the edge. (See Optical on the Up: OFC 2018.)
There's a lot of interest in a technology called Time Sensitive Networking where you can specify, on a service flow by service flow basis, how the network will treat [traffic] in terms of total latency. So you can have some high priority and low priority traffic and network will behave differently depending on the types of traffic.
The customer conversations have been very broad and deep – a lot of discussions on connectivity and capacity but also discussions on "How do I build the network so it can adapt to all these different traffic types I'm seeing come on board?"
KZ: How does Ciena's Adaptive Network come in to play in terms of 5G planning?
SA: The Adaptive Network is a principle that says we'll have telemetry and analytics fed up out of the network from both physical and virtual devices. It's going to go through a layer of intelligence and automation and based on observables in the network, ideas around self-learning and machine learning, being able to go back and program the network to behave or operate the way you want it to and adapt to certain circumstances. It's a way you can do all the capacity and connectivity planning necessary for 5G but in a way that the network itself can respond to changes.
If you think about how networks are built, there are always functions that are inherently automatic in the network -- a good example is protection switching. When something breaks or fails, you want it to automatically protect. Some functions in the network are inherently autonomous, like mesh restoration. "Plug-and-play" is generally autonomous, IP re-route is autonomous. But those things are always governed by some rules or policies. Where the adaptation piece comes in is when the network is presented with things that are unanticipated -- additional traffic, connectivity and service types. How does that adapt that combination of automatic and autonomous functions, physical and virtual devices? How does all that adapt and reconfigure to meet the needs of what the network is being asked to do?
That's why we like the term "adaptive" instead of "autonomous" -- we believe the entire network -- the virtual and physical elements, algorithms, the way the network is stitched together -- that entire ecosystem needs to be adaptive.
Autonomous networks to some people are like Skynet from the Terminator movies -- they don't want it to be autonomous, they want it to be more policy or intent-based and give it high-level instructions and the network should be intelligent enough to figure out what to do.
— Kelsey Kusterer Ziser, Senior Editor, Light Reading
This is an edited version of a story that was originally published on Telco Transformation's sister site, Light Reading. To see the full story, click here.