Idle power leakage detection by cluster and service
Detect and quantify persistent idle power leakage at cluster/service granularity, enabling remediation and compliance reporting. Per Helios IdlePowerLeakageDetectionInput schema.
Unique identifier for the data center facility
Unique identifier for the cluster from scheduler/DCIM
Service/job/workload identifier mapped by scheduler
ISO8601 UTC timestamp. Freshness requirement: < 5 min latency
Per-cluster or per-service instantaneous power draw
Percent resource usage (CPU/GPU/memory)
True = within scheduled maintenance, mask data for window
Current SLO/SLA status as JSON pointer or policy document ref
Incident or alert log reference (CSV/JSON)
Resources below this utilization are considered idle
Upload facility specs, equipment data, or operational logs to improve analysis accuracy
Upload Context Documents
Drag and drop files or click to browse
Supported: TXT, CSV, JSON, Markdown, PDF, Excel (max 10MB each)