How grafana/beyla works
a. Input Flow:
• Application Processes → Process Discovery
• Network Traffic → eBPF Probes
• Kubernetes API → Service Discovery
b. Processing Flow:
• eBPF Events → Protocol Analyzers
• Protocol Data → Data Transformers
• Transformed Data → Exporters
c. Output Flow:
• Metrics → Prometheus Endpoint
• Traces → OpenTelemetry Collector
• Kubernetes Metadata → Enrichment Pipeline
Key Architectural Elements
a. eBPF Subsystems:
• kprobes/uprobes Handlers
• Network Traffic Capture
• SSL/TLS Interception
b. Language-Specific Instrumentation:
• Go Runtime Integration
• Java/JVM Support
• Node.js/Python/Ruby Support
c. Security & Configuration:
• Privilege Management
• Configuration Management
• Resource Controls
External Integrations
a. Monitoring Systems:
• Grafana
• Prometheus
• Tempo (Traces)
• Mimir (Metrics)
b. Container Platforms:
• Kubernetes
• Docker
• Container Runtimes
Implementation Details
a. Core Libraries (show as modules):
• BPF Module (/bpf directory)
• Protocol Handlers (/pkg/internal/ebpf)
• Exporters (/pkg/export)
b. Support Systems:
• Configuration Management
• Logging & Debugging
• Health Monitoring
Deployment Contexts (show as separate zones)
a. Host Deployment:
• Direct Process Instrumentation
• System-wide Monitoring
b. Kubernetes Deployment:
• DaemonSet Mode
• Sidecar Mode
• Cache Service
The Kubernetes integration capabilities
Remember to:
• Keep the core eBPF functionality central
• Clear data flow paths
• Security boundaries
• Configuration points
• Version dependencies
• Scalability aspects
This design reflects Beyla's architecture as an advanced observability tool that leverages eBPF for automatic instrumentation while maintaining flexibility in deployment and export options.