Preface: Art Fewell is a well-known blogger in the SDN community and also pens a regular column on Network World. Art has been having interesting video conversations with various thought leaders in the SDN space and we’ve collaborated with him to make available these conversations as part of our Contributed Articles program here on SDxCentral.
In this conversation, Open Daylight Chair and Brocade Chief Scientist David Meyer catches up with Art Fewell on Open Networking TV to discuss software-defined intelligence and the impact of machine learning on networks and compute.
For the full experience, watch the video below, or read the transcript. For our TL;DR readers, here’s a couple of quick snippets of the conversation, especially as it pertains to networking:
- Machine learning is exploding right now. With SDN, the network has become programmable. The SDN phase has given people abilities that weren’t there before. Now, with machine learning, how can we make the network more intelligent? How close are we to having machines write their own codes and become more intelligent than humans? Maybe human intelligence isn’t the best intelligence after all.
- Machine learning takes data and a machine learning algorithm and outputs a program you hope generalizes the cases you haven’t seen yet. The better the data, the better your program. You have to train the neural network to learn just like the human brain.
- When you think about how your intelligence works, you’ve got all these sensors; eyes, ears, tactile, etc. What that’s doing is throwing a big stream of data at your neocortex which is then building statistical models that allow you to understand the world around you. The network is doing the same thing.
- Part of what makes this possible is the way the network was built. You don’t have to code up all these hard algorithms yourself these days. You can just get them which makes access to this technology easier.
- Today, everything we do is data oriented. Big data, it’s everywhere. The way we process this is with machine learning. In the overlay and the underlay there’s all these counters or telemetry in the network. If you think about looking at those in a vertical way, if you just take a cut through the overlay & underlay and look at all of the data as sort of a one data element then you have all of those going through time. Are there patterns that are in that data set? Sure. We just can’t see them. As humans we can’t see because we aren’t that good at patterns. We are good at pattern recognition but not at that scale. So what does this mean for humanity?
In the video below, David shares more wisdom on the future of AI and the impact AI will have on networking, the data center and humankind.