The Dashboard (module rail → Dashboard) answers one question at a glance: is my platform healthy, and who is using it? Its sidebar groups three areas: Overview, Usage and Cluster Manager.
Overview
The Overview presents live metrics for the compute cluster:
| Metric | What it tells you |
|---|---|
| Total Nodes | How many servers are in the cluster. |
| Total GPUs | Accelerator capacity serving models. |
| Avg Utilization | How busy the GPUs are on average. High sustained utilisation means demand is near capacity; very low suggests spare headroom. |
| Avg Temperature | Average GPU temperature — thermal health of the hardware. |
| Power Draw | Current power consumption of the accelerators. |
| Total Consumption | Energy used over the viewing window. |
| Cluster Average | The cluster-wide rollup of utilisation trends. |
Trends matter more than single values: utilisation creeping up week-on-week is your cue to plan capacity.
Usage
The Usage area breaks down activity and cost — requests, token consumption and spend over a selectable date range, per user or across the workspace.
- Every member sees their own usage (also reachable from the account menu → Usage).
- Administrators see workspace-wide activity for capacity and chargeback decisions.
Full guide: Usage & Credits.
Cluster Manager
The operational console for the infrastructure: node status, container status, storage status, and cluster configuration (adding nodes, console access).
Full guide: Cluster Manager.
Good routines
| Cadence | Check | Where |
|---|---|---|
| Daily | Node/storage health | Cluster Manager |
| Daily | GPU utilisation vs capacity | Overview / GPU Monitoring |
| Weekly | Usage trends and top consumers | Usage |
| Monthly | Capacity planning inputs for hardware decisions | Overview + Usage history |
Related
- GPU Monitoring — detailed per-GPU view
- Usage & Credits
- Cluster Manager