Autonomous & Self-Healing AI Networks: Helping Universities to Scale Smarter

Automatic basic network troubleshooting enables institutions to shift Level 1 basic support to AI tools, decreasing the need for hands-on involvement from help desk staff and improving response times. Automated networking also brings efficiencies to the deployment process.

“We had a customer deploy about 200 access points in a couple of hours using templatized configurations,” says Aaron. “They even had students come in and do it. They plugged it in, scanned it, and it was up and running. They only needed to download the right configurations. The more automation you have, it just makes life easier for staff. They don’t want to spend their time having to go plug in access points and configure them. That’s a big job.”

Automation Increases Network Reliability

Offloading routine network configurations to AI not only improves efficiency but also enhances the reliability of the campus networks.

“Humans making config changes during business hours occasionally leads to service interruptions,” explains Ward. “Anyone who’s ever configured a network switch by hand knows the pain of accidentally removing all VLANs from an uplink port — it’s a rite of passage.”

Automated configuration not only prevents typos, but solutions like Mist also offer an automatic rollback feature that restores a device’s configuration to the last known working state if a config change takes it offline, according to Ward.

Network reliability gets an added boost from AI self-healing features, which help universities manage the network bandwidth more effectively across campus.

“Bandwidth is key. Campus bandwidth usage is something that can really get taxed,” says Jackson. “The network infrastructure that supports all the work on campus is routinely handling heavy workloads. Built-in automation can help throttle unauthorized bandwidth usage. So if network thresholds are reached, AI can simply reboot a router and dump some of those unauthorized connections, which allows you to recoup that bandwidth for authorized campus needs.”

DIG DEEPER: This AI Playbook provides broader higher ed AI strategies.

Challenges on the Road to a Self-Driving Network

While the benefits are clear, careful planning and a number of other factors must be considered before moving toward a fully autonomous networking environment.

“Institutions often face cost barriers when adopting new technology, whether they use a software-based solution or another hardware-based solution,” says Jackson. “Also, the timeline can be difficult because you want to avoid downtime during the school sessions to be able to implement and test your new setup. And it can be difficult to have the right expertise in place to deploy the solution.”

IT leaders must also assess what they are comfortable handing off to AI. Some network automations are easier to adopt than others.

“We let customers choose which actions go into self-driving mode,” says Aaron. “There are about 10 common actions, like rebooting a stuck port. AI can do that reliably. Why do you need a human to do it?”

Aaron lists a misconfigured VLAN or noncompliant software code among other examples of common self-driving options.

“The main actions are downloading the compliance software,” he added. “So these are some of the activities that can move more toward autonomous today, and we’ll be adding more.”

LEARN MORE: How a network assessment can uncover optimizing opportunities.

One of the biggest challenges of introducing autonomous features to the network, though, isn’t the tech — it can be the IT services culture itself.

“Many engineers, including myself, have built a career on networking a certain way,” Ward says. “We’ve toiled countless hours and cut our teeth doing things the ‘hard way,’ often with technology that hasn’t changed in over a decade. There is safety and comfort in keeping things unchanged.”

His advice? Start small.

“Changing how a team operates requires time, training and understanding,” he added. “Rolling out self-driving and automated networks is a long process but has immense value in the end. Start small and grow as you better understand and become more comfortable with the new technology.”