putilov_denis - stock.adobe.com
In 2022, AI network management is all about growing trust
AI network management tools are poised for significant growth in the coming year, but making the most of the tech's capabilities will require trust.
Anthony Wild is the global network operations manager at Johnsonville Sausage, the largest sausage brand by revenue in the U.S. He's one of three people who manage the Wisconsin-based company's network of more than 30,000 nodes spread across offices, manufacturing facilities, and personal employee devices.
Networking has transitioned from monolithic platforms with known and profiled devices to tens of thousands of unknown endpoints, Wild said. Managing that much complexity requires advanced technology that didn't exist several years ago.
"I look at the way we did business three years ago, writing traditional IP-based access lists, and it's impossible [today]," Wild said. "We just can't do business that [old] way anymore."
Wild is not alone. Networks worldwide are increasingly complex, and the talent pool of qualified IT staff is not keeping pace. Many companies have implemented AI-driven network management tools to address the problem -- in Wild's case, Cisco AI Network Analytics.
"Being able to maintain our headcount here and not have to grow our IT staff as our business grows and has more acquisitions -- that's been critical [to reducing expenses]," Wild said.
As networks shift into the cloud, more data than ever is available to train AI in resolving connectivity problems. As a result, AI-driven network management tools have gained steam across the industry. Factors including the increasing maturation of the technology and more complex networks required for a work-from-home environment have contributed to a boom year in 2021 for network AIOps.
Major developments in 2021
Prominent vendors injected AI into network management through the acquisition of startups. Many 2021 releases focused on adding intelligence to more parts of the network -- SD-WAN, Wi-Fi access points, switches -- with the goal of end-to-end network visibility.
They also released AI-driven automation products and continued to build out the multi-cloud networking environments that are critical in producing the network data needed to train AI. In 2021, significant developments in AI/machine learning network management included:
- Juniper Networks integrating its cloud-based Mist AI into the 128 Technology SD-WAN that the company acquired in 2020. Mist ingests telemetry data to detect network problems and maintain adequate bandwidth for applications.
- Arista advancing its CloudVision management console to deliver automation, telemetry and analytics across the data center and campus networks.
- Cisco integrating internet intelligence technology ThousandEyes with Catalyst switches and the AppDynamics application performance monitor. The integration provides network and application visibility stretching from the campus or branch to SaaS applications and software running on public or private clouds.
- Extreme Networks launching a beta version of ExtremeCloud IQ CoPilot, a subscription-based tool to deliver AI insights for IT network administrators.
- And Aruba, a Hewlett Packard Enterprise company, adding auto-fixes to the AI Insights The platform, which was previously capable of predicting or identifying network problems, can now self-heal a selection of less problem-solving intensive network issues, such as balancing the number of devices connected to a 2.4 GHz vs. a 5 GHz wireless band.
The state of AI in networking today
AI-powered features in network management tools typically include technology that notifies IT staff of network problems, recommends fixes and makes changes approved by IT staff. Some products can fully automate problem detection and resolution and send information about the issue to the IT team after the fact.
Roseville Joint Union High School District in California uses Aruba's Edge Services Platform to help manage a network that serves more than 10,000 students and 600 teachers, plus support staff. The district has found that the tips and suggestions the technology provides have helped to reduce time spent troubleshooting.
Bob LaliberteAnalyst, Enterprise Strategy Group
"It's almost like somebody is always there watching [the network] and letting us know if there are things [going wrong]," said Dave Todd, senior network administrator. While Todd would like to automate some network tasks eventually, he doesn't trust the AI to resolve network issues independently quite yet.
Todd isn't alone.
"[The AI] technology isn't necessarily the hard part; the hard part's the cultural aspect," said Bob Laliberte, an analyst at Enterprise Strategy Group (ESG). "People in IT tend to be very conservative because their jobs depend on it … how do you get someone comfortable with technology that takes decisions out of their hands?"
Where is AI network management going?
For Wild, getting comfortable with using AI to drive automation of the Johnsonville Sausage network is mostly just going to take more time.
"It's definitely a shift in mindset. We're used to doing things kind of the old-fashioned way … dictating our own policy, and writing things statically," Wild said.
Many enterprises will be going through that change in thinking in 2022, when ESG expects companies to start trusting AI to perform routine network chores and automatically make some fixes.
"They're getting more comfortable with the technology," Laliberte said. "When I'm talking to AI vendors in the network management space, I'm always asking them, what's that time to get comfortable? It varies from organization to organization, but the time tends to be in weeks to months."
Once organizations begin allowing an element of automation into their networks, the benefits they'll see will be significant, said IDC analyst Mark Leary.
Using AI in networking will "free up [IT] staff so they can do something of higher value," Leary said. He gave examples of researching new network technologies, bringing more converged security and cloud elements into the network, and strategizing about better aligning network capabilities with business goals.
"If you're a staff person, that's going to translate to much more job satisfaction, greater job growth, and certainly from an organizational perspective, better retention," Leary said.
Enterprises can dip their toes into AI-driven network automation by putting only select functions into their mechanical hands, Laliberte said. Popular candidates include provisioning devices, discovering IT assets and performing lifecycle management tasks like patching and upgrading.
"AI [in network management] is coming; it's not going away," Laliberte said. "For anyone who's avoiding it, 2022 is the time to start looking at learning about it, bringing it in, testing it, embracing it."
Enterprise Strategy Group is a division of TechTarget.
Madelaine Millar is a news writer covering network technology at TechTarget. She has previously written about science and technology for MIT's Lincoln Laboratory and the Khoury College of Computer Science, as well as covering community news for Boston Globe Media.