The phrase “generative AI” has been inescapable over the past 18 months, with business leaders trying to leverage artificial intelligence solutions like ChatGPT for everything from building new software applications to drafting their marketing plans. But Brandon Butler, an IDC research manager covering enterprise networks, says that the current hype cycle will also lead to increased adoption of AI for other applications, including managing and optimizing IT networks.
“At IDC, we believe we’re entering the era of AI everywhere,” says Butler.
“We’ve seen a lot of buzz about generative AI recently, but that is the precursor to wider adoption of AI across the IT industry. The coming years are going to be defined by AI, but we’re going to see the focus broaden across all parts of the IT stack, and that includes applying AI to manage network infrastructure better.”
Here are four ways that Butler says organizations can improve their enterprise networks with the help of non-generative AI plus one bonus use of generative AI for network management.
Enhanced Analytics and Insight Generation
Increasingly, Butler noted, IT vendors are incorporating AI into their offerings to help organizations better understand what is happening in their networks. These AI features can quickly learn the normal behavior and traffic patterns of an enterprise network, and then identify abnormal behavior that might indicate performance or security issues. One important use case of AI network analysis, Butler said, is the automated updating of firmware for networking infrastructure.
“That may not sound super exciting, but if you have an enterprise network with dozens of access points or controllers or switches – and each one of those is running a different firmware – these AI tools can perform discovery and then execute updates without impacting the user experience.”
Other analysis use cases for the network include client/device profiling, personalized or industry-specific service-level baselining, and identifying and quantifying the risk posed by particular users, devices and applications.
Automated Solutions and Swift Responses
“When there are problems in the network, organizations want to reduce the time to fix those problems by as much as possible,” Butler said.
“A machine learning or AI platform can actually help the enterprise to understand where problems are coming from and the best way to fix them.”
This means that AI tools can not only rapidly diagnose problems through automated root cause analysis (RCA), but can then also resolve those issues through guided remediation or even automated resolution. As a result, organizations can speed up their response times, reduce costly network downtime, and lessen the workload on their IT staffers, allowing them to focus on more strategic tasks.
Optimizing Performance, Continuous Improvement
According to IDC’s 2023 “Future of Connectedness” survey, 33 percent of business and IT leaders say optimizing network performance will be a top network management improvement stemming from machine learning and artificial intelligence algorithms, more than any other outcome. (Only 2 percent say AI and ML will not improve network management.)
Broadly speaking, this means increasing network utilization via data-driven insights and operational recommendations based on historical data patterns. Butler points to tools that can make real-time assessments of wide-area networks, determine the capacity of different links into a central site, and then dynamically shift from one path to another to optimize traffic flow.
“These tools are using ML and AI capabilities to understand past traffic, predict future traffic, and optimize a WAN to provide a high-quality user experience without someone actually having to go in and turn dials and push buttons to enable it,” Butler said.
Anticipating Challenges and Opportunities
Finally, Butler says AI tools are helping network administrators better predict future network behavior. Beyond the in-the-moment insights already mentioned, Butler says, AI can provide longer-term recommendations based on historical patterns and a network’s existing state of play.
“Based off of what we’ve seen historically from a network, and what’s happening with it now, these tools can predict that the network is going to see a huge spike in application or user traffic at some specific point in the future,” Butler said. “Then, organizations can use that information to prepare their networks.”
By giving network administrators the information they need to proactively prevent problems from occurring, AI features can help organizations ensure that users and applications have the capacity they need to achieve peak performance.
Generative AI for Network Management
While Butler emphasizes the opportunities for AI applications beyond generative technologies, he notes that organizations can also use large language models like ChatGPT to improve and streamline their network management practices. Specifically, vendors and enterprises can feed information about their network management into an AI solution, providing users with the chat-based, conversational experience they have become accustomed to receiving.
“We’ve seen some vendors upload all of their networking documentation into a private ChatGPT, if you will,” Butler said. “Then you can ask that tool how to solve problems in your network.”
Although Butler sees the current AI conversation is largely centered on future-looking generative applications for creative work, he said existing network management tools can help organizations solve their IT problems right now.
“Enterprises are just starting to wrap their heads around how they can use these features,” he said.
“It’s an exciting time to think about how organizations can use AI, and how that’s going to continue to evolve into the future.”
