NAF AutoCon 2 Recap

It’s been a minute since my blog post, so I apologize – life has been busy over the past year. Coincidentally, my last post recapped AutoCon 0, so I guess it makes sense I come back from my blogging hiatus to talk about my experience at AutoCon 2!

Summary

AutoCon 2 was special for me because it was my first opportunity to present a conference talk! I’m very grateful to Scott and Chris (co-founders of the conference) for giving me the opportunity. I’ve recorded podcasts and webinars, but never spoke in front of a room of hundreds of people. It was an exhilarating experience – I’ll dive deeper into my preparation and overall feelings about my presentation later in this post. Besides speaking, I was impressed with how much the conference has grown. I don’t have exact stats, but I heard there were over 550 people registered for the conference and somewhere around 25 booths. It was sold out almost a month before the conference! In addition, there were multiple workshop tracks (also sold out) that allowed you to get hands-on learning from network automation gurus/industry experts. I didn’t make it to the workshops this year, but I heard they were well worth it! Now let’s get into some of the conference highlights!

Conference Highlights

AutoCon once again did not disappoint. It’s a community-focused conference that aims to bring people from all over together in one room to solve real-world problems – and you can feel the enthusiasm and passion! There were practitioners sharing their automation success stories and experiences on the main stage and vendors showing off their network/infrastructure automation solutions in the expo hall, but the real value comes from the hallway conversations you saw happening throughout the week. People from different companies/industries meeting for the first time bonding over a funny networking story or the different methods they’ve used to get buy-in for network automation initiatives at their respective company. There’s something special about getting together in-person and collectively sharing experiences and struggles when striving for a shared goal – in this case, adopting network automation.

Before moving on, I wanted to highlight a few of the vendors and talks at the conference.

Vendors/Products

There were many vendor booths in the expo hall, and I could write a whole blog about all the cool products I saw, but I wanted to focus here on a few vendors in particular: NetBox Labs, Arista, Itential, and Ops Mill. Let’s dive into each one!

NetBox Labs

NetBox Labs is the commercial steward of the popular network source of truth product NetBox. At the conference, NetBox Labs announced NetBox Discovery and NetBox Assurance products. NetBox Discovery helps engineers more easily gather and ingest network data into NetBox. NetBox Assurance helps engineers identify operational drift from the intended state defined in NetBox. NetBox has a bright future and I can’t wait to see how NetBox Discovery and NetBox Assurance elevates the product further.

Arista

Ever since AutoCon 0, Arista’s automation efforts around Arista Validated Designs (AVD) and Arista Network Test Automation (ANTA) Framework have been on my radar. Over the past year, I’ve been using Arista cEOS in containerlab to recreate network topologies for dev/testing purposes. If you haven’t tried containerlab, you’re missing out. I highly recommend checking out the containerlab docs and get started today! This year, my good friend JulioPDX showed me a demo on how you can use AVD to configure a containerlab network (Arista spine/leaf) topology and use ANTA to run validation tests against each device. It really inspired me to look further into ANTA and try it out myself. Thanks again Julio!

Itential

Back in May, Itential introduced a new product called torero. At first, I didn’t understand the problem it was solving (hence their blog post title – “introducing torero >_ the automation gateway you didn’t know you needed”). It just seemed like a CLI tool that helped abstract some of the nitty gritty details of executing Python code, OpenTofu plans, or Ansible playbooks. I thought, “Why not just learn the basics of Python or the Ansible CLI and run the playbook yourself?!”. However, it’s a little more than that. It really helps streamline the process to operationalize scripts and playbooks. It helps increase adoption of automation scripts written by devs by allowing non-programmers to use the scripts without knowing the intricacies of setting up an execution environment. For example, how long would it take for you to explain how a Python virtual environment works if the person has never installed Python? Or how do you explain a Python Package Index (like PyPI) and installing Python packages using pip? I think a tool like torero really helps promote the message “not everyone needs to be a developer/programmer”, which is 100% true. An entire network team doesn’t need to learn Python or Go or name-your-language in order to adopt network automation. Obviously, there needs to be a developer or someone who understands code on the team in order to develop, troubleshoot, and enhance existing automation, but there’s a distinction between the different personas: Those who develop automation and those who are users (or superusers) of automation. Superusers may be a special group of users that know enough about programming to setup a dev environment and make changes to existing code, but doesn’t necessarily feel comfortable developing code from scratch. I plan to continue watching the progress of torero, as I think there’s a place in the industry for such a tool.

Ops Mill

The last vendor I wanted to highlight is Ops Mill and their flagship product Infrahub. I’ve known Damien Garros, their co-founder and CEO, for awhile and have enjoyed watching him grow this company with his other co-founder, Raphael Maunier. One of their biggest recent announcements was the release of Infrahub 1.0. Infrahub is an interesting product and competes with other network source of truth products such as NetBox and Nautobot. However, what sets Infrahub apart is the freedom to define your own data schema. The freedom can feel relieving but also stressful. If you don’t have a strong idea of how you’re going to organize your data, it may feel overwhelming. Ops Mill is doing a good job curbing some of the stress by developing a schema library, which should help streamline and standardize infrastructure-related data structures. The other major feature is the native integration of version control in its graph database. By having version control baked into the database, it provides the opportunity for branching/merging and continuous integration (CI) directly in the platform. The company and product are still young, but I can see how this product will continue drawing attention from companies looking for a network source of truth.

From the Main Stage

The main theme for many of the conference talks this year, to no surprise, was AI. I’m personally still getting a grip around AI so I tried attending any talk that discussed AI. So far, I’ve determined network operations (NetOps) is a huge opportunity for AI/ML and probably the best place to start implementing it. This shouldn’t be a surprise, as it seems like most people I talk to agree NetOps is a great place to start incorporating AI, and it makes sense when you think about it – a vast amount of data from multiple sources that requires quick analysis/correlation when network issues arise. Another interesting concept that was presented was Natural Language Processing (NLP). NLP essentially allows users to interact with a chatbot using natural human language. This allows network engineers to quickly find out how the network is operating without logging into multiple systems to parse syslog data, review monitoring alerts, search access control logs, etc. I encourage you to check out all the presentation talks when they are uploaded to the Network Automation Forum YouTube channel, but the talks that focused around AI that I found very interesting were (not in a specific order):

  • Jeremy Schulman – AI Driven Advanced Network Observability
  • Du’An S. Lightfoot – Network Whisperer: Building the Ultimate AI Network Agent
  • Phillip Gervasi – The human factor of using LLMs in network operations

Jeremy, Du’An, and Phil are all phenomenal speakers, which helped deliver their messages effectively. Like I said, these talks were specifically of interest to me because I’m interested in learning about AI and how it can make network engineers lives easier. Check out the NAF YouTube channel soon to watch all the presentation talks! Speaking of talks, let’s dive into my first experience presenting a conference talk.

My First Conference Talk

I wanted to take some time to give insight into my talk. More specifically, how I came up with the topic, how I organized my thoughts into a presentation, and ultimately how I felt presenting the talk.

Back in January 2024, when the call for speakers opened for AutoCon 2, I was in the middle of writing a Cisco Press book about Cisco pyATS with John Capobianco. I’ve been writing the book with John since July 2023, trying to knock out a chapter every week or two. In addition to writing the book, I was also working on some interesting projects at work where I was writing tests to validate the operational state of the network after maintenance window changes. So given I was writing a book about a network automation testing framework and working on projects that incorporated network testing, I thought it would be worth a shot to submit a talk about network testing and the importance of beginning with a minimal risk task (like network testing) when beginning your network automation journey. To be clear, I’ve been promoting the benefits of network testing since I began working with Cisco pyATS years ago (around 2019). Cisco pyATS was the first network testing framework I worked with. I’ve worked with testing frameworks such as unittest and pytest for software testing, but not to run tests against the network.

When I received an email from Chris and Scott that my proposal was selected for AutoCon 2, I was ecstatic! I’ve recorded many YouTube livestreams/videos, webinars, podcasts, and tech sessions, but I’ve never presented a conference talk. As part of my preparation and organizing my thoughts into a presentation, I wanted to tell a story to keep the audience engaged. The problem is network testing isn’t exactly the most exciting domain in the network automation space, but I believe it’s one of the most important! Here’s the basic flow I wanted to establish for my presentation:

  1. Define the problem statement – Relate the subject of my talk (network testing) back to the main question of the conference – “Why haven’t we seen full adoption of network automation, yet?”.
  2. Ensure we are speaking the same language – Define the different types of tests (i.e. unit testing, integration testing, E2E testing) and testing framework components (fixtures, parametrization, markers, testscripts, etc.).
  3. Tooling examples – Provide examples of tools and Python libraries we can use to collect data from the network and integrate with specific network testing frameworks.
  4. Success stories – Show how I’ve used some of the previously mentioned tools and testing concepts to successfully execute real-life projects.

The success stories are the most important, but providing the necessary background information about the different tooling and testing frameworks are required to provide proper context.

I’ll admit, while building my slides and thinking through my presentation, I became nervous and started second-guessing myself. I started asking questions such as, “Should I even talk about this tool/concept?”, “I definitely put too many words in my slides.”, “Is one meme enough?”. Some points were valid, but I ultimately moved on and focused on the delivery and presentation of the material. I’ve never practiced for any other presentation as much as I did for this talk. I thought through what I was going to say for each slide without just reading from the slides. I focused on using the slides as reference points and not to describe my entire thought process. This helped out a lot during practice because I didn’t feel pressured to reiterate every bullet point mentioned in the slide – some bullet points were just implicit. Once I got to Denver, I felt prepared due to the consistent practicing and I felt confident in the material I was covering, which helped calm my nerves. Once it was time to present, I was ready. Many people I met at the conference leading up to my presentation wished me good luck and it made me feel like they were on my side (not that I expected people to be against me). The words of encouragement were helpful and I felt like I delivered my talk effectively. If you were at the conference and caught my talk, feel free to reach out and give me your feedback – I’m always looking to improve!

Conference Takeaways

After AutoCon 0, I remember feeling the excitement of just having a group of like-minded individuals in one room to discuss their experiences and struggles with adopting network automation. Now at AutoCon 2, where the number of attendees and sponsors basically doubled from AutoCon 0, I still feel the same excitement. The vibe is still the same. The problems have changed and it seems like conversations about automation have matured, but everyone is still working together, as a community, to try to understand what’s holding us back from adopting network automation.

Last year at AutoCon 0, there seemed to be a focus on understanding why we haven’t seen adoption. Some of the answers included getting buy-in from your organization, starting with low-hanging fruit and getting easy wins, and being patient with the journey (because adopting any new process/culture change is a journey). This year, the focus changed to how adoption is being achieved. Some presenters talked about automation they’ve built in-house to solve business problems, some talked about how you can take advantage of AI and use it to enhance existing automation, and there was one talk by William Collins, I encourage everyone to watch, where he discusses different channels you can leverage to get business buy-in.

Wrapping Up

If you made it this far, thank you! This blog post became way longer than I expected, but I really wanted to get all my thoughts down. I hope you were able to take something away from this post and it encourages you to attend the next AutoCon conference, either in Europe in the spring (AutoCon 3) or in the US next fall (AutoCon 4). If you have any questions or feedback, please feel free to leave a comment or hit me up on Twitter/X (@devnetdan) or on LinkedIn. Thank you again for checking out this post. Until next time!

2 thoughts on “NAF AutoCon 2 Recap

Leave a comment