Observability Overview
VION provides a built-in observability stack that collects metrics, logs, and traces from your edge gateways and makes them available through Grafana dashboards.
Architecture
Data Flow
- Edge components (Dale runtime, Mesh gateway) emit telemetry via the OpenTelemetry Protocol (OTLP)
- The Dale runtime also emits per-actor vitals (
vion.actor.*) alongside its logs and metrics, feeding the Actor Vitals dashboard - OTel Collector running on the gateway receives, batches, and forwards telemetry to VION Cloud
- Mimir stores metrics (Prometheus-compatible) and Loki stores logs
- Grafana provides dashboards to query and visualize the data
What Data Is Available
| Data Type | Source | Storage | Query Language |
|---|---|---|---|
| Metrics | Edge-gateway resource usage (memory, CPU, GC, messaging) | Mimir | PromQL |
| Actor Vitals | Per-actor health from the Dale runtime (vion.actor.*) | Mimir | PromQL |
| Logs | Structured logs from all edge components | Loki | LogQL |
Tenant Isolation
All telemetry data is tagged with a tenant_id label. Each tenant can only see their own data — isolation is enforced at the Grafana level through teams and folders.