Monitor resource utilization¶
The Monitoring > Resource monitoring tab provides visibility into resource utilization metrics for a Workload, helping you monitor performance, identify bottlenecks, and understand autoscaling behavior. Use this tab to evaluate resource usage, navigate trade-offs between speed and cost, and ensure the Workload efficiently uses its allocated resources.
To access Resource monitoring, select a Workload from the Deployed Workloads list and then click Monitoring > Resource monitoring. The tab displays summary tiles showing aggregated and current values for key metrics, along with an interactive chart that visualizes resource utilization over time.
Resource utilization summary tiles¶
The Resource monitoring tab displays summary tiles at the top of the page, showing both the aggregated value over the selected timespan (the primary value) and the Live value at the time of the request. The aggregated value represents the average value over the selected timespan. Each tile also shows the timestamp of the most recent update. Click a metric tile to update the paired chart to display that metric over time.
The following metrics are displayed as summary tiles:
| Metric | Description |
|---|---|
| Replicas | Active replicas the platform currently observes from runtime metrics—not the configured replicaCount in Workload settings. After a scale-up, replacement, or autoscaling event, this value can lag until pods are ready and reporting. |
| CPU Utilization / GPU Utilization | The percentage of CPU or GPU cores being used across the Workload's replicas. |
| Memory Usage | The amount of memory being used across the Workload's replicas. Displayed in bytes or appropriate units (KB, MB, GB) based on the scale of usage. |
Configured replica count
To see how many replicas are requested, check Settings or GET /workloads/{workload_id}/settings (Runtime settings). The Replicas tile shows what is running now.
Resource utilization chart¶
The Resource monitoring tab displays an interactive chart that visualizes the selected metric over time, helping you identify patterns and understand resource consumption trends.
The chart displays the selected metric over time, with the following elements:
| Chart element | Description |
|---|---|
| Time (X-axis) | Displays the time represented by each data point, based on the selected resolution. |
| Metric value (Y-axis) | Displays the replica count (cardinality) or the average value of CPU, GPU, or memory for each time period. |
To view additional information on the chart, hover over a data point to see the time range and metric value.
You can configure the Resource monitoring dashboard to focus on specific time frames. The following controls are available:
| Control | Description |
|---|---|
| Artifacts | Updates the dashboard to show resource utilization for the selected artifact. Historical stats are scoped to that artifact, making it easy to compare resource usage before and after a replacement. |
| Range (UTC) | Sets the date range displayed for the Workload date slider. You can also drag the date slider to set the range. |
| Resolution | Sets the time granularity of the Workload date slider. Available options are Minutes, Hourly, Daily, and Weekly. |
| Refresh | Initiates an on-demand update of the dashboard with new data. |
On the CPU Utilization chart, GPU Utilization chart, and Memory Usage chart, use the Replicas dropdown to filter the chart to show resource utilization for one or more specific replicas.
