Artificial Intelligence & Machine Learning: SD-WAN is Evolving

by Yulia Duryea
April 16, 2018

Machine Learning and AI Promise to Take SD-WAN Into the World of Intent

Last month, my mother-in-law’s best friend came to town, so she rounded up “the gals” for dinner and drinks. A night without the kids is rare for me (and significantly more relaxing) so I found myself in the midst of half a dozen 60 to 70-year-old women. The conversation eventually got to technology; how different and difficult it is for their generation to embrace it (though all had smartphones in their pockets). They’ve noticed facial recognition on Facebook; same for police cameras. One lady going to France next month raved about Google translate. Another nonchalantly mentioned a recent ride in a self-driving Tesla.

AI and Machine Learning: SD-WAN is Evolving

As tech savvy as they seemed, I didn’t dare try to explain how artificial intelligence and machine learning play a role in all of these things (or in what I do in my chosen profession). Though both have most definitely started to touch our daily lives, no matter what your age.

What changed? How did my mother-in-law and her friends start benefitting from AI without knowing it? First, compute power is now big enough for the amount of data that needs to be gathered, sorted, calculated and stored to make AI happen. Second, neuro-networks only became real rather recently, and we have since shown that we can train computers via deep (machine) learning. And finally, steady advances in enterprise networking have made it possible to conduct real time transactions with cloud-based AI solutions to put all that data to work on an increasingly broad basis.

Intent based networking on the mainstream horizon

Network transformation has progressed in leaps and bounds, but it’s really only just begun. Our solutions architecture team has been actively evaluating intent-based network technology and we’re close to moving ahead. A few years ago, the advent of SD-WAN made separating the control and routing planes possible. Windstream SD-WAN Concierge allows customers to set their own routing policies, based on what SLAs are more important to their unique business needs (availability, jitter, latency, etc.). Providing our customers’ unprecedented network visibility and control, such as identifying which applications are using the most bandwidth, and prioritizing how that needs to change to deliver against business and customer experience objectives.

Now imagine if instead of asking customers to define how much jitter/latency is acceptable on circuit 1 prior to switching traffic to circuit 2, we asked them to more broadly define their ultimate intent. How much bandwidth does each site need? As well as when, and under what conditions? An intent based system intelligently identifies the appropriate parameters and configurations, and automatically re-routes their traffic. Truly zero touch this time, giving customers visibility without exposing too much complexity.

The future of networks is closer than you think

Today, our industry leading fully-managed SD-WAN Concierge solution has the intelligence to tell network/IT leaders which applications use the most of their bandwidth. Along with the power to make changes to how they access bandwidth based on that data. Tomorrow, with the addition of intent based network functionality, we will also be telling them where their network investment dollars went, and will automagically re-route their traffic to meet their intent. IT leaders (and those who determine IT budgets) want to know how much money their cloud applications are costing them every minute. And they want to use that information to influence that spend. It won’t be long until they can – the network as a platform has never been closer.

At that point, our network becomes a huge sensor, comprised of a large multi-dimensional data set with the ability to detect anomalies. A lot of interesting and hard to see relationships live in multi-dimensional data. So, every time any network element is misconfigured (which can happen on occasion) our network learns from it. Over time, it makes the best decisions given all the data it has collected, processed and computed. This is predictive analytics; the creation of SD-WAN policies based on machine learning.

This, fellow futurists, is the network and digital transformation our engineers at WE have embarked on. And we are closer than ever to delivering it!

Now if only I can figure out a way to explain this to my mother-in-law and her friends…