Observability (local)

What to monitor on-site without relying on cloud.

IntelFactor supports fully local observability.

Metrics to monitor

  • Inference latency (p50/p95/p99).

  • FPS and frame drops vs line speed.

  • GPU utilization.

  • Queue depth and buffer growth.

  • Disk utilization (evidence partition).

  • Camera health heartbeat.

  • Defect rate drift vs baseline.

  • RCA response time (when triggered).

Expected outcome

Operators and QA can see station health, shift summaries, and RCA recommendations without cloud connectivity.

Last updated