In 1993, a group of network professionals came together in San Diego for the first-ever NANOG event. Fast-forward to last week, and the North American Network Operators’ Group — now an established staple in the networking community — gathered for its 93rd meeting in Atlanta, February 3-5, at the Atlanta Marriott Marquis.

Between keynote speakers, hackathons, socials, and tutorials, NANOG’s three-day sprint went without a hitch, continuing to serve as a space for networking, sharing knowledge, and discussing the latest advancements in internet technology.

“There’s no other event like this in the industry,” said Jonathan Black, NANOG’s executive director. “NANOG provides an accessible and collaborative environment where Internet professionals converge to shape the future of networking.”

James Ashton, Vice President of Network Operations at 365 Data Centers, told us, “NANOG is always a good place to catch up on what is going on in the industry that we don’t see in our normal day-to-day.”

NANOG holds three conferences per year in different cities all across North America. The last one, its 92nd event, was in Toronto, and it’s already gearing up for its 94th in Denver in mid-June, with a registration deadline of May 11 for those who are interested.

Among NANOG 93’s attendees were a diverse bunch, including enthusiasts, industry leaders, small business owners, and service providers. There were also several representatives from major companies like Google Cloud, Cisco, and Amazon Web Services.

Highlights also included talks on AI’s environmental, social, and technological impacts, with sessions on data center optics, test network labs, and IP inventory management. Undoubtedly, the hackathon and social events — trivia, pickleball, and a run club — rounded out the week.

For attendees, the value lies in being able to shed the weight and responsibilities of everyday life to get creative, insightful, and have fun among like-minded folk. Tucker Preston, Gcore’s head of edge connectivity and peering, is one of the many who continue to participate.

“I had a fantastic time at NANOG 93 in Atlanta this week,” said Preston. “In a digital-first world, there’s still no match for bringing like-minded individuals and groups of people together in-person to exchange ideas, network and discuss shared concerns, opportunities and goals.”

The event centered on how AI is crucial for managing GPU traffic, optimizing model partitioning, and ensuring smooth data flow across networks, according to Preston.

“With AI workloads becoming increasingly latency-sensitive, edge deployments and optimized architectures will be key to balancing efficiency, scalability, and performance in the AI-driven enterprise,” he added.

Network automation means more power is needed. Better flexibility, security, and real-time monitoring must be prioritized because, as networks become more automated, they need to handle changes quickly.

Preston suggested that as AI adoption accelerates, networks should prioritize scalable inference computing and optimized infrastructure to keep up with demand.

“While most open-source models can run efficiently on a handful of GPUs, larger inference workloads require well-orchestrated clusters and low-latency networking solutions,” he said.

One topic that particularly stood out for 365 Data Centers — both internally and in discussions at NANOG — was the evolution of lab and testing tools, where Ashton agrees that automation is the key to streamlining network operations.

“Containerlab and some of the other automation tools will help us streamline our current processes a good deal and hopefully make some of our lab work happen at a much faster pace,” said Ashton.

Beyond automation, 365 Data Centers plans to strengthen some of its partnerships surrounding network security and data analytics that were present at NANOG.

“We are excited for updates and new features in the Kentik Flow and DoS mitigation platform,” Ashton said. “Our partnership there has been very fruitful, and their new advances look to give us a lot more usable data than we’ve ever had access to in the past.”

A few notable experts in attendance included Cat Gurinsky of Apple, who participated in a panel on the future of network automation. To name a few more, Mauricio Rojas and Alejandro Guevara from Nokia led a tutorial on leveraging KRM for declarative network automation, and Chad Lamb of XKL presented a lightning talk on the evolution of optical transceivers from 10G to 800G.

Randy Zhang of Cisco discussed using AI agents for automated networking script development, while Sharada Yeluri presented on networking for AI. There were also workshops on network automation and modern approaches to designing, deploying, and testing network labs. This event’s Women in Tech panel featured key voices like Yolandi Cloete from DE-CIX and Bhawna Jain and Erin Geiger from Arelion.

Two keynote speakers were the main events of the three-day-long conference: Scott Robohn and Yun Freund.

Scott Robohn

On Monday, Robohn took the stage with his presentation, “Building Community around Network Automation.”

As the co-founder of the Network Automation Forum (NAF) and founder of the Total Network Operations (TNOps) project and podcast, Robohn’s expertise lies in his more than 30 years in the industry.

His keynote was in the form of a Q&A, focused on network automation — which he admits has been a bit slow on the uptake, but there’s no shortage of those eager to push its adoption.

Robohn highlighted the community’s growth and shared ways organizations can collaborate to continue driving network automation.

Yun Freund

Freund, the vice president of Google Cloud, Global Networking Technology, also held a keynote titled “AI-Powered Network 5.0 — A Paradigm Shift” on Tuesday.

If her name sounds familiar, it may be because she was named one of 2022-2024’s “Top 50 Women in Tech Leaders to Watch” and was also recognized as a “Woman of Influence” by the “Silicon Valley Business Journal” in 2019.

In her Q&A keynote, Freund discussed how AI/ML advancements are shaping the future of global networks.

She discussed the growing complexity of network design as hyperscalers grow, and how traditional, rigid networking approaches can’t keep up, pointing to Network 5.0 as a solution.