SeaGL 2024 | Zero-instrumentation observability based on eBPF

Observability transforms raw system data into actionable insights, empowering teams to detect anomalies and optimize performance. While essential for modern infrastructure, traditional approaches often struggle with:

📉 Implementation complexity requiring code instrumentation

⏳ Resource-intensive maintenance across distributed systems

🔍 Data overload without clear troubleshooting pathways

eBPF-Powered Telemetry Collection
This talk demonstrates how extended Berkeley Packet Filter (eBPF) redefines observability by:

🌐 Capturing Container Intelligence
Real-time network call mapping (HTTP/gRPC/TCP)

Filesystem operation tracking (reads/writes/modifications)

Process execution monitoring across namespaces

📊 Unified Data Pipeline
Metrics: Kernel-level performance indicators

Logs: Context-rich event correlations

Traces: Cross-service dependency mapping

From Data to Action: Smart Troubleshooting

Discover how to:
✅ Identify latency spikes through auto-correlated traces
✅ Detect configuration drift via filesystem change timelines
✅ Pinpoint network bottlenecks using protocol-level metrics
✅ Reduce MTTR (Mean Time to Resolution) by 60-80%

🔮 Attendees will gain actionable strategies to implement lightweight, production-safe observability that scales with Kubernetes-native environments.

Perfect for DevOps engineers and SREs seeking to reduce monitoring overhead while improving system reliability.

Event link: https://seagl.org/archive/2024/zero-instrumentation-observability