KubeCon NA 2023: What a Conference!

Now that we’ve had time to decompress from KubeCon, we wanted to do a writeup about our collective experience. Six of us spoke at the conference and Charity participated in a panel, so we included short talk recaps. 

Key highlights

• The talks were a hit! Purvi’s browser talk and my observability during development talk especially seemed to garner excitement. People approached us throughout the week to say how impactful the talks were—no follow-ups, no asks, just the urge to tell us how great our talks were. That was cool! 
• Meeting and talking to other OTel folks in real life was great. It was an awesome opportunity to interact on a real human level.

Talk recaps

We’ll try to keep these short since there were so many!

Build in Observability While Developing - Jamie Danielson: Every developer needs to see what they’re doing. I wrote a blog about this talk since it contained a lot of good info—check it out!

A Practical Guide to Debugging Browser Performance with OpenTelemetry - Purvi Kanal: What’s crucial in website performance today, and how can OpenTelemetry help? Purvi wrote a short recap about her talk.

Journey to Becoming an OpenTelemetry Approver - Jamie Danielson: Go from a beginner developer to a recognized Open Source contributor, with me as your guide!

OpenTelemetry Tracing for Monoliths - Phillip Carter: Phillip gives practical advice to increase your understanding of software systems—no Kubernetes or microservices architecture required.

Observing a Large Language Model in Production - Phillip Carter: In this talk, Phillip explains how we instrumented Query Assistant, what we tracked, what SLOs we set up, and how we measured our improvements as we iterated on the feature. 

Dynamic Sampling in Practice - Kent Quirk: Control the costs of top-quality traces with dynamic sampling. This is not a trivial undertaking, and Kent is one of the top experts in the world, so listen to his take!

OTTL Me Why Transforming Telemetry in the OpenTelemetry Collector Just Got Better - Tyler Helmuth, Honeycomb & Evan Bradley, Dynatrace: At a high level, this talk covers the OpenTelemetry Collector, why you'd use it, and the problems that OTTL solves. Following a real-life Kubernetes Event situation, you'll see how OTTL enables manipulating a log in a way no other component in the Collector can. You'll also learn about other OTTL use cases and the future of OTTL in the Collector.

Detailed timeline

Monday

Monday morning started with some breakfast from a nearby cafe. We walked to the conference center to watch Tyler and Kent’s presentations, both of which were very well done. Tyler and Evan really showed their expertise and made a complex implementation seem simple. We were inspired by Kent’s conversational style and contagious enthusiasm.

Monday was a major focus day: Purvi and I stayed at the AirBnB and ordered lunch in, so that we could practice and refine our talks.

Tuesday

The big day! Purvi gave an amazing presentation on browser observability, complete with polish from the prior day’s practice sessions, and even punched it up with a shiny new layout. I was blown away, and could tell others in the audience also loved the talk.

After that, I was a bundle of nerves and stressing about my upcoming talks. Vera, Faith, and Purvi came to my aid and saved the day with their suggestions on how to move things around. While the nerves were still there, I at least felt more confident in the final product by that point.

I delivered my talk to a packed room, and it went off without a hitch. Charity—even though she participated in a panel in the session before mine—was in the front row, cheering me on. Overall, my presentation was well received, and by the time I finished with Q&A, I was ready to take a nap—but I had another talk to give in just over an hour! 

I furiously jotted down a ton of speaker notes into my slides so I had a guide to keep me grounded. Funny enough, we had a couple technical issues: the timer never started, so I had no idea how much time had elapsed or how long the talk was going. The mic also had feedback throughout the session, and at one point, a speaker fell over and everyone (myself included) looked around to see what the commotion was about. Phillip’s talk on LLMs happened at the same time so I unfortunately didn’t get to attend, but from what I heard, it was super insightful and got great feedback.

With the conference day done, I happily followed some fellow Honeycombers around the Kubecrawl event, enjoying a couple beers and chatting with folks.

Wednesday

Contribfest seemed to go over well with the attendees. A few OTel folks were able to talk about what went well and what could work for next time. For example, sending out some recommended prep work, like cloning repos and installing packages, would save some time at the start. 

There was a Dungeons and Deployments talk right before dinner, which was a fun and clever way to talk about Kubernetes in the context of D&D. 

Thursday

We sat front row for Phillip’s Tracing for Monoliths talk, which was packed full and contained some great insights. Thursday was a short day, so after some wandering around booths and picking up some swag, a group of us went on a boat architecture tour. This was a really nice—albeit cold—way to learn about and appreciate Chicago history, and I’m glad we went.

Friday

I spent Friday mostly decompressing and getting to know other bees better. We started with a hearty breakfast at Yolk. A group of us chatted about LLM observability for about an hour, and then moved on to board games. We all walked to an early dinner at Fatpour, which allowed for more bonding. The whole day was excellent, even though we were all pretty tired from a long week.

See you next year!

All in all, KubeCon was an exciting conference where I learned a lot, met so many interesting people, and got to talk about topics that are important to me. I’d call KubeCon NA 2023 a huge win, and I can’t wait to go next year!