Thousands of KubeCon + CloudNativecon North America attendees braved cold rain – and even snow – to attend 16 co-located events in the Salt Palace Convention Center in Salt Lake City. With talks aimed at every level from beginner to expert covering a broad swath of cloud native technologies including AI, observability, eBPF, WebAssembly, and Istio, the always popular co-located events were filled with enthusiastic participants. 

Here’s a slice of how the day unfolded. Want to follow KubeCon + CloudNativeCon North America 2024 in real time? We’re on Instagram, Linkedin, X (for KubeCon), and X (for CNCF).

Platform Engineering Day

The advent of internal developer platforms promises to revolutionize every step of the development process and the excitement around IDPs was the focus of the Platform Engineering Day co-located presentations this year. In Portals and platforms, two Ps in a pod? How great interfaces make for good operability, Abby Bangser, principal engineer, Syntasso, and Jorge Lainfiesta, reliability advocate, Rootly, talked about the universal challenges around creating development templates that can grow, scale, and adapt to future needs. They also discussed the equally tough hurdle of getting developers to actually use those templates once they were created.

To ensure long term success, build templates that have “operability” in mind but also keep other “ilities” on the list including maintainability, usability, and extensibility. Maintainability is achieved by looking at the past, while usability is how to allow for present adoption and extensibility will ensure developers can build something in the future they can’t even imagine today.

Wondering what your team should be demanding of your platform and tooling? Compare efforts with the CNCF Platform Engineering Model.

Cloud Native + Kubernetes AI Day

Artificial intelligence can challenge everything we’ve come to believe about cloud native, and Autumn Moulder, director of infrastructure & security at Cohere, tackled that head on in her presentation, From Supercomputing to serving: a case study delivering Cloud Native Foundation models.

Moulder offered her company’s story – taking an ML infrastructure from one cloud to multi-cloud and many clusters in less than six months – as object lessons in how to take cloud native learnings and apply them in the AI world. 

Five challenges made the company stronger, she explained. Cohere had to overcome GPU capacity constraints, GPU high failure rates, the reality of tightly coupled software and hardware, complex multi-team deployments, and managing allocation and utilization. Her team had to get creative and was particularly successful using Kueue.

Cloud Native Startup Fest

To kick off an afternoon targeted at entrepreneurs, Kelsey Hightower, retired distinguished engineer and Megan Reynolds, VC at Vertex Ventures, hosted a fireside chat, Startup resilience in a post-ZIRP (zero interest rate policy world). To get things rolling, Hightower suggested founders ask some hard questions beginning with something super basic:

Does the product work? To answer this, have a customer try it out without any input at all; if they can’t get it up and running, it’s clear the product doesn’t work. Also, it’s important to be careful with the term customer, Hightower said. The word needs to be reserved for those who’ve paid for the product, he emphasized. Also, he said he’s a big believer in early stage companies taking field trips to potential customers and really immersing themselves in the business. Ask those potential customers what you should build, and how much they’d pay for it. And, of course, look to open source because it’s got a proven track record as a great place to start new businesses.

Istio Day

For those wondering what it would be like to finally embrace service mesh, Confluent shared their experience during Confluent’s service mesh journey – building security and reliability one sidecar at a time, with Adam Sayah, Solo.io and Cody Ray, Confluent.

The driving force behind Confluence’s decision to move to service mesh was a combination of factors, starting with the compliance fed ramp, as well as an internal push for zero trust, Ray explained. The company had a lot of moving parts to deal with including tens of thousands of Kubernetes clusters, hundreds of services and almost 5,000 connections that needed continuous security monitoring.

With such vast scale – and a need for increased productivity and security – Ray felt the time was right to make the case for service mesh, and his ROI calculations showed an expected savings of $3.5 million + per year, not to mention centralized observability, improved compliance and faster delivery.

Watch the highlight reel from Day 0 of KubeCon + CloudNativeCon North America 2024!

Get the first look at some of the fun, networking, and learning that went on today!