The Long Arc of Network Automation: Lessons from a 25-Year Journey

Summarizing Jason Edelman's closing keynote from AutoCon3

"I've been fortunate enough to be part of two careers in one," reflected Jason Edelman as he opened his closing keynote. The first half—traditional networking, routing, switching, certifications. The second—network automation, APIs, and open source. His journey mirrors that of an entire industry grappling with transformation.

Standing before 650+ attendees at AutoCon 3, Edelman delivered a retrospective that was part history lesson, part state of the industry, and part vision for what's ahead. For someone who got his CCNA in May 2001 and witnessed the evolution from CLI to API to AI, his perspective offers unique insights into where we've been—and where we might be going.

The Template Trigger

Edelman's automation awakening came from an unlikely source: the word "template." Working on campus network migrations in the early 2000s, he'd constantly ask to see the templates his colleagues used for new site deployments.

"It was always a Word document," he recalled. "Ten pages, fifty pages, or more. IP addressing schemes, VLAN allocations, step-by-step procedures." The absurdity struck him: "There was no automation happening. I knew deep down there was a better way, and no one was talking about it."

This resonates with anyone who's lived through the era of configuration by documentation—detailed procedures that humans executed manually, step by step, hoping not to make mistakes.

The OpenFlow Catalyst

The industry inflection point came around 2011-2012 with OpenFlow and software-defined networking. While SDN didn't transform overnight as many expected, it served as a crucial catalyst.

"I literally thought overnight was going to change," Edelman admitted. "But enterprises take time for refreshes. Nothing changes overnight." However, the venture capital investment, startup activity, and thought leadership around SDN created momentum that ultimately included automation, APIs, and operations—concepts that weren't prominently featured in early SDN discussions.

The Community Foundation

Edelman dedicated significant time to recognizing the people and projects that built today's automation ecosystem:

Pioneering People: Kirk Byers (NetMiko), David Barroso (NAPALM), Jeremy Schulman (Juniper PyEZ), and Peter Sprygada, whose move from OpenStack to Ansible was "pivotal for the industry" in legitimizing networking automation for enterprise adoption.

Foundational Projects: NetMiko ("you can't do network automation today if you haven't at least tried NetMiko"), NAPALM, Ansible's networking modules, and Cisco's DevNet program (over 500,000 people trained).

Recent Innovation: Container Lab as the standout project of recent years, accelerating both automation and networking learning through accessible virtualization.

His recognition of these contributors wasn't mere nostalgia—it was acknowledgment that current automation capabilities rest on years of community-driven innovation.

The Current State: Power Tools and Data

Today's automation landscape centers on what Edelman calls "power tools"—Python scripts and Ansible playbooks that solve immediate problems. These represent the most common entry point for practitioners.

"You might have taken one of Kirk's Python classes and you're saying, 'I just learned this and I want to apply it to rotate community strings across 100 devices.'" This problem-solving approach creates tangible outcomes that justify investment and demonstrate value.

The evolution from personal power tools to organizational self-service represents a critical transition: "It's about going from a team of one who's building and executing automation to thinking about how that's distributed to maybe their team, their organization, and maybe even exposed as higher-level services."

The Data Revolution

Perhaps the most significant shift has been the recognition of data as fundamental to automation success. "The vast majority of talks this week mentioned data or source of truth," Edelman observed.

The emergence of NetBox (2017), followed by Nautobot and ForHub, created healthy competition that "makes us all better." Having choice allows organizations to make decisions appropriate for their specific problems and businesses.

However, Edelman noted that vocabulary remains challenging: "Source of truth is one of those topics" where interpretations vary. Understanding context and intent matters more than precise definitions.

The AI Inflection Point

On artificial intelligence, Edelman struck a balanced tone: "I do think the time is now to dive in and explore." However, he cautioned against overnight transformation expectations, having learned from the SDN era.

"We're seeing the art of the possible," he explained. "I don't think we have the frameworks, products, or toolchains yet that will give us mass adoption. But when these tools get adoption, it's going to be at a faster rate than we saw with network automation."

He demonstrated AI's potential with vibe coding—using voice to generate automation scripts: "When I think about the ability to build a power tool or prototype, it just seems like it makes sense to experiment and learn."

The Skills Evolution

One of Edelman's more provocative points addressed the evolving baseline skills for network engineers. While avoiding the "should network engineers learn Python" debate, he outlined practical realities:

"When we think about baseline skills for the next generation—Linux, Git, a little Python, making an API call—even if you're buying tools, there are APIs. Can you test by making an API call in Postman? Can you open an IDE, import requests, make an API call?"

This isn't about becoming a programmer, but about adapting to a world where APIs are ubiquitous and basic automation capabilities become table stakes.

The Future Landscape

Looking ahead, Edelman identified several trends:

Open Source Continuation: "Open source is going to continue to play a pivotal role," likely with AI integration into existing projects or new AI-focused networking tools.

Self-Service Evolution: Marketplaces, freemium models, and reduced sales friction as engineers resist lengthy procurement processes.

Enterprise AI Adoption: Will take time but accelerate rapidly when it arrives, similar to network automation patterns.

Skills Stratification: Continued growth between "haves and have-nots" in technical capabilities, with baseline expectations rising.

The Community Challenge

With 650+ attendees at AutoCon 3, Edelman challenged the community: "What's the next project? Who wants to write the next book? Everyone in here really has something to share."

The growth from small meetups to major conferences demonstrates momentum, but also creates responsibility. How do we continue expanding the community while maintaining accessibility for different experience levels?

The Abstraction Debate

The session concluded with a fascinating exchange between Edelman and Dinesh Dutt (former CTO of Cumulus Networks). Dutt argued that Cumulus was "too far ahead" with abstraction—they could do what people use multiple tools for today, but the concept was too advanced for adoption.

"Cumulus is nowhere today," Dutt observed. "The abstraction was way too advanced for people to grasp." He warned against "analysis paralysis" and the danger of "eloquent" solutions (SDN controllers, AI) that sound compelling but miss practical adoption needs.

This tension—between visionary abstraction and practical adoption—remains central to automation evolution.

The Persistence of CLI

Despite decades of API development, CLI remains prevalent. Edelman's perspective: "I'm actually really thankful for the last 25 years of CLI networking. It made my career, got me great jobs, helped me meet great people."

Rather than viewing CLI as legacy burden, he sees it as foundation for understanding problems that automation solves. "Living that pain, living that hell, helps understand the problems we're going to solve."

Why This Matters

Edelman's keynote captured the network automation community at an inflection point. The foundational work is done—tools exist, communities are established, and early adopters have demonstrated value. But mass adoption remains elusive.

The challenge isn't technical anymore. It's cultural, organizational, and educational. How do we help more organizations and individuals navigate the journey from CLI-centric operations to automation-enabled efficiency?

The answer lies in understanding that everyone's journey is different—different learning styles, organizational appetites for risk, and technical starting points. Success requires meeting people where they are while providing clear paths forward.

As Edelman concluded with words he lives by personally: "It begins with a dream. We have to dream that there is a better way. Once we dream that, we have to believe we can go do it. Once we have that belief, it's the fun part—we have to go do it and make it happen."

Twenty-five years from CLI to API to IDE represents remarkable progress. The next twenty-five years, shaped by AI and community growth, promise even more dramatic transformation. But as Edelman's journey demonstrates, lasting change happens through persistent effort, community building, and the willingness to keep dreaming of better ways to work.


Chris Grundemann

Executive advisor. Specializing in network infrastructure strategy and how to leverage your network to the greatest possible business advantage through technological and cultural transformation.

https://www.khadgaconsulting.com/
Previous
Previous

The Black Box Problem: Why Trust is the Missing Piece in Network Automation

Next
Next

Beyond Data: Why Design-Driven Automation is the Missing Link