Getty Images/iStockphoto
Latest AI network trends signal future of network automation
AI will revolutionize all aspects of network operations, from management to security. Experts discussed AI's potential and the challenges of readiness at ONUG Spring 2024.
AI will influence all areas of network operations, from network management and monitoring to security and traffic analysis.
The ONUG Spring 2024 conference in Dallas focused entirely on the advent of AI. Networking experts, engineers and enthusiasts shared presentations focused on the AI boom expected to hit the networking landscape soon. Each presentation varied in topic and scope, but common themes about AI network trends recurred throughout the event.
Experts largely attributed the general interest in AI to an increase of data volume. Network professionals struggle to manage data located across distributed network environments, accessed by users in various locations. Now, they consider how AI can benefit their businesses, primarily through automation. However, concerns about AI readiness suggest AI needs to reach a sufficient level of maturation before broad deployment.
Large volume of data in network operations
Many presentations at the conference echoed how the increased volume of applications, data and information poses challenges to network professionals. For example, organizations now have thousands of applications distributed across multiple on-premises and cloud infrastructures.
IBM, in particular, has upward of 3,000 distributed applications that serve 170 countries, said Andrew Coward, general manager of software-defined networking at IBM.
"The network wasn't designed for the way applications are working today," Coward said. In his presentation, Coward said networks should be rearchitected in ways that prioritize applications across an infrastructure.
Network professionals face more issues than application overload, however. They also have to contend with influxes of data and system alerts that have grown too complex to manage.
"The biggest problem we have is the collection of data," said Mark Berly, CTO of data center networking at HPE's Aruba, in a separate presentation focused on how AI enhances network traffic analytics. "I have to understand [which devices are] talking, what ports they're talking on and the duration of their conversations to make intelligent decisions."
Network security professionals experience similar issues as they work to secure their environments. In a panel discussion about AI in network security management, Craig Matsumoto, senior analyst at S&P Global Market Intelligence, cited a report from 451 Research in which half of the respondents said they can't respond to alerts they receive from security systems.
"Twenty-nine percent said three-quarters of the alerts were things they just weren't able to handle, and this is a matter of volume," Matsumoto said.
AI for data management and operations automation
Network professionals grapple with the large amount of data and information thrown at them, which has contributed to the increased interest in AI. Experts said automation specifically, as part of an AI tool, can help teams manage operations.
Network automation tools are already available on the market, but automated AI tools have the ability to analyze data and autonomously create predictions based on the provided information. In terms of network traffic analysis, Berly said AI will be used more for augmentation and analytics, rather than as a sentient tool that makes decisions for network professionals.
Ali Shaikh, chief product officer at Graphiant, agreed: "That logic [is] pushed to the edge at the source of where the data lives and can infer what it needs to do with the traffic flow."
Network security tools with AI operate similarly. Ameet Naik, director of product marketing at Zscaler, said static tools aren't sufficient to secure network environments. AI-driven automation tools can help network security professionals with policy management, he said.
AI not ready yet, despite hype
Experts also recognized that further advancements of both AI and the network need to occur before broad deployment. For example, Coward said, before AI can integrate into networks, organizations need to have sufficient data for the tool to make data-driven decisions.
"[Make] sure the network is instrumented correctly and tools for provisioning and management are automated," he said. "If you don't have that capability, AI will not be able to solve any problems."
In terms of edge AI, Shaikh said networks need to be rearchitected before AI can infer information from data located at the edge. The network infrastructure itself also needs to continue to develop at different levels of the OSI model. As this progression occurs, network professionals can work to further understand the use cases for automated traffic management.
"We're going to be in the augmented automation analytical space, learning the use cases [and] developing as we go," Shaikh said.
Deanna Darah is associate site editor for TechTarget's Networking site. She began editing and writing at TechTarget after graduating from the University of Massachusetts Lowell in 2021.
