Open Source Summit North America 2025

Our cloud native world has become more than just tooling, it's an entire ecosystem with many add-ons, complementary tools, when it comes to K8s CRDs, and services that provide its powerful capabilities and infinite scale...but at what cost?

In this talk, we'll share first of its kind research that will highlight the 5 most common OSS cloud native tools killing your observability costs. We'll start by exploring how different observability tools structure pricing, the complexities that compound cost calculation, and especially which OSS tools in your stack are the most resource-intensive services.

You'll discover how you can know whether it's KEDA or Karpenter, ArgoCD or Kyverno ballooning budgets. But don't panic! We'll wrap up with good practices for configuring popular tools to be more economical, so you can leverage the powerful K8s ecosystem without breaking the bank.

Monty Python’s Black Knight is the opponent that couldn’t lose. Even after all of his limbs were cut off, he offered a draw: “it’s just a scratch.”

FoundationDB (FDB) is a distributed transactional key-value store that is very difficult to defeat just like the Black Knight. Open-sourced in 2018 after an acquisition by Apple, FDB was designed to be a common layer: almost all databases have a backing key-value store. Many have built on top of it including Snowflake, Adobe, & Datadog.

FDB got it right: transactions, distributed by default, and extreme reliability. Kyle Kingsbury (aphyr) the author of Jepsen series on distributed systems correctness, said: "haven't tested foundation in part because their testing appears to be waaaay more rigorous than mine."

We demo a live FDB cluster and try to disrupt its operations. Our attempts are informed by real world experience supporting a metadata service for billions of objects globally.

When we finally succeed, we show how backups and disaster recovery resurrect FDB. We’ll learn about highly resilient design patterns and operations. We have battle scars, and want to help others!

Traefik is one of the most popular open-source projects in the world, with over 3 billion downloads and a top 15 spot on DockerHub. As a powerful Ingress and Gateway Controller, Traefik simplifies exposing, securing, and managing services and APIs dynamically and at scale—whether in simple setups or complex cloud-native environments.

Just one year after the release of Traefik v3, we're already taking things to the next level! In this session, Emile Vauge (Traefik Creator) and Nicolas Mengin (Traefik Maintainer) will unveil the exciting new features coming in Traefik v4, including:
- A new plugin system for even greater extensibility
- Pre-routing operations to optimize traffic handling
- Enhanced TLS certificate management for better security and automation
- Improved configuration management for a smoother experience
- … and much more!

Join us to get a sneak peek at what’s next for Traefik and see how these innovations will make your cloud-native journey even easier.

Containers have changed how we build and run services. The days of FTPing a binary up to a server are gone, because our platforms expect to run containers. We build container images at every stage of development, whenever we want to test our services, and when we deploy them.

If container images are what we need, could our development tools help us build them? Yes, they can! In this talk, we'll take a container image apart, see what makes it tick, then put it back together again from first principles - all using Swift!

Swift is a high performance, memory-safe language which is ideal for server-side development. We will:

* download a container image, take it apart by hand and explore what’s inside;
* cross-compile a Swift service effortlessly to different Linux distributions, on x86 or ARM, statically or dynamically linked, from development environments on macOS or Linux;
* use Swift's pluggable build system to produce container images efficiently and automatically for every build;
* test the image.

Containers are a universal building block of modern services. Even if you're not yet using Swift, these ideas and principles also underpin your current build and deployment workflow.

Context propagation is the cornerstone of observability in distributed systems, but traditional approaches often falter in non-linear workflows like message queues, event-driven architectures, state transitions, or shared resources. Span links powered by OpenTelemetry, bridge this gap by enabling connections between spans across unrelated execution contexts.

This session explores a practical use case where span links augment context propagation in an event-driven microservices system. We'll demonstrate how to track a single user's transaction across services that communicate asynchronously. Using tools like OpenTelemetry and compatible backends, we'll show how span links resolve visibility challenges, uncover hidden latencies, and maintain trace continuity even when the standard parent-child relationships break.