Limitations
This page documents known limits of the VION observability stack, including data retention, ingestion rates, and data filtering.
Data Retention
Metrics and logs are retained for a limited period. Retention is configured per environment and may vary. Contact VION support for the exact retention period in your environment.
Ingestion Limits
| Limit | Value |
|---|---|
| OTLP batch size | 512 events per batch |
| Batch timeout | 5 seconds |
| Collector memory | 64 MiB hard limit per gateway |
| MQTT connections | 32 concurrent connections per gateway |
If the telemetry collector reaches its memory limit, it will start dropping data to prevent the gateway from becoming unresponsive.
Queue and Buffering
Telemetry data is buffered on the edge gateway when VION Cloud is temporarily unreachable:
- Queue storage: File-backed, up to ~115 GB
- Retry strategy: Exponential backoff (1s → 30s)
- Compaction: Automatic, every 15 seconds if threshold exceeded
Data is not lost during short cloud outages — it queues on the gateway and syncs when connectivity is restored.
Filtered Data
To reduce noise, certain high-frequency MQTT messages are excluded from tracing by default:
- Property state updates (
property/state) - Measuring point state updates (
measuringPoint/state)
These can be enabled by setting trace.enabled=true in the telemetry configuration if needed for debugging.
Tenant Isolation
- Each tenant sees only their own data (filtered by
tenant_id) - Cross-tenant queries are not possible
- Custom dashboards are scoped to your tenant's folder
- Viewer role cannot modify pre-built dashboards
What Is NOT Collected
- Personal data (PII) — no user-identifying information is sent to the observability stack
- Raw MQTT payloads — message contents are not logged by default
- Historical property values — use the Cloud API measuring points endpoint for time-series property data