May 13 2024

What Is AI-Native Networking, and How Can It Benefit Higher Ed Institutions?

Network troubleshooting is faster and easier when AI can pinpoint the source of issues and, in some cases, fix problems on its own.

With lean IT departments and an ever-increasing number of devices on campus, higher education is a logical environment in which to deploy artificial intelligence in networking. Increased visibility means faster troubleshooting, stronger security and a better user experience — all of which translates to a lighter load for IT. Campus users increasingly expect institutions to deliver seamless, reliable connectivity with fast and easy problem resolution. AI helps to make that possible, even for IT teams managing thousands of devices.

EdTech recently spoke with Tom Wilburn, the vice president of global enterprise sales at Juniper Networks, to discuss AI-native networking and how it can support strategic improvements in higher education technology.

DISCOVER: Juniper Networks solutions for enhancing higher ed networks.

EDTECH: How is AI enabling IT professionals to manage and, especially, to troubleshoot networks?

WILBURN: Traditionally, if the entire network went down, IT knew about it because everybody was disconnected. But now, if someone experiences a poor connection on a Zoom call, IT won’t know unless that user complains, and then IT has to figure out the problem. We believed, when we started on this problem a decade ago, that if we could get enough telemetry and if we had a tool like AI to make sense of all that data, we could tell you that a user was experiencing a problem related to the network. That’s what we set out to do: leverage AI as a next-gen form of automation.

Now, when we determine that a user is having a bad experience, we start capturing packets automatically, with no human intervention. AI can correlate massive amounts of data and, in seconds, come to a good conclusion about 80 percent of the time. In many cases, AI can correct the issue automatically. Suppose AI determines that 10 students connected to the same access point are having a problem and the issue is a bad Ethernet cable. The network can’t fix a bad wire, but it can alert IT and say, “If these users haven’t called you yet, they’re going to, and here’s the problem.” This can be a huge time-saver.

Click the banner below to learn how to optimize device management in higher education.


EDTECH: How does a cloud-managed architecture enable AI-native networks to leverage greater telemetry?

WILBURN: For years, institutions have relied on premises-based, wireless LAN controllers. We’ve moved that “brain” into the cloud and now provide it as a service. The data plane stays local, but the management and control plane move to the cloud, enabling a rapid response no matter how many access points are deployed. Similarly, the network management console can be a challenge and an expense. With the cloud, all you need is a browser, and we can scale up resources to ensure the network is responsive. The cloud also enables us to handle the telemetry we get: 150 states from every device, every minute. AI is only as good as the data it acts on, and the old wireless LAN controllers don’t have enough CPU or storage to handle the amount of data we need to leverage AI effectively.

Recently, we added a digital twin to our AI engine: Marvis Minis, a riff on Iron Man’s AI assistant J.A.R.V.I.S. Imagine an IT staffer makes a mistake inputting a new security rule at 1:30 a.m. Typically, no one would discover that until the next morning when everyone comes to class and the network is down. Marvis Minis is the equivalent of thousands of users on every wireless AP out there, so at 1:31 a.m., when that rule starts to block traffic it wasn’t supposed to, we’d have 11,000 Minis say, “Wait, there’s a problem.”

EDTECH: How can AI help institutions bolster cybersecurity?

WILBURN: Our solutions have integrated network access control, compatible with eduroam, to automatically determine whether someone is university-affiliated and should be allowed to access the network. We also built in a wireless intrusion detection system to detect and disable any rogue APs through which a bad actor might try to connect to the network.

EDTECH: What misconceptions do you think people have about AI-enabled networks?

WILBURN: Even for IT professionals, it’s hard to believe AI can make this big of a difference. Institutions may want a network upgrade but lack the budget to replace the entire thing, so they might worry about doing a partial upgrade to AI and managing two different systems. But what I hear from customers is that even if they deploy AI only partially, the visibility they gain is such a significant improvement that it’s worth it. Once they see how powerful AI can be, they typically continue to upgrade until they’ve brought their entire network up to the same level, even if they have to do it over more than one budget cycle.

Our customers are reducing their mean time to resolution by 90 percent. With AI, we hope to eliminate at least half of their trouble tickets; in some cases, we’ve eliminated more than 90 percent. IT departments that adopt these solutions generally find that they receive a lot fewer complaints, and users are happier.

Brought to you by:

imaginima/Getty Images

Learn from Your Peers

What can you glean about security from other IT pros? Check out new CDW research and insight from our experts.