Pixie is a Kubernetes-specific, real-time debugging platform that collects metrics, traces, profiles, full request and response bodies, and network data without manual instrumentation or application redeployment. It uses eBPF to gather fine-grained telemetry and provides queryable, scriptable visualizations through PxL. section 6.2.2 section 9.1.3
A practical use is answering, “What is going on in my Kubernetes cluster right now?” Users can inspect application profiles in a pod or across a node, query live data, and build dashboards for operational investigation. Telemetry stays in memory and local to the cluster; OpenTelemetry export supports longer-term retention, because historical data is outside Pixie’s intended scope. section 6.2.2 section 9.1.3
Operationally, Pixie combines:
- eBPF-based collection without code changes or redeploys. section 6.2.2
- PxL queries inspired by Python Pandas and fully scriptable dashboards. section 6.2.2
- Rich live signals balanced against memory storage and cluster-local scope. section 6.2.2 section 9.1.3
