Zero-instrumentation Observability Based on eBPF

How do you build reliable observability across your infrastructure without the overhead of instrumenting every application? This session explores how eBPF is changing the way teams collect and use telemetry data in containerized environments.

Observability is essential for detecting and resolving issues quickly, but the traditional approach is costly. Incorporating instrumentation across every application is time-intensive and resource-heavy, creating barriers that slow teams down before they've gained any real visibility.

We dive into how eBPF enables the collection of metrics, logs, and traces at the kernel level, covering how it tracks container activity including network calls and filesystem operations, all without the usual instrumentation burden.

The session closes with practical guidance on putting that telemetry data to work, walking through how teams can use these signals effectively for troubleshooting distributed systems and reducing time to resolution.

If you're an SRE or DevOps engineer looking to simplify your observability stack while gaining deeper visibility into your infrastructure, this talk offers a grounded look at what eBPF-powered telemetry collection makes possible in practice.