Observability Stack

Narwhal provides full-stack telemetry integrating metrics, logs, distributed traces, and network flows into a single unified operational dashboard.

Telemetry Pipeline

[ Kubernetes Pods & Nodes ]
      │
      ├─ Metrics (Prometheus Exporters) ──► Prometheus ──► Grafana
      ├─ Logs (Alloy Agent) ──────────────► Loki ────────► Grafana
      ├─ Traces (OpenTelemetry SDK) ─────► Tempo ───────► Grafana
      └─ Network Flows (eBPF BPF Maps) ──► Hubble UI

4 Observability Pillars

1. Metrics (Prometheus Operator): Node CPU/memory, XFS storage consumption, and APISIX latency alerts
2. Logs (Grafana Loki + Alloy): Label-indexed container log aggregation and streaming
3. Traces (Grafana Tempo): OpenTelemetry distributed trace correlation with logs via TraceIDs
4. Network Telemetry (Cilium Hubble): Real-time service topology, DNS latency, and packet drops via eBPF