Kubernetes

The Kubernetes view breaks down container infrastructure costs across your clusters and identifies workloads that are over-provisioned or at risk of throttling. Open it from Lenses > Kubernetes in the left sidebar.

Use Kubernetes when you need to:

• Understand how CPU, memory, storage, and network costs break down across your clusters.
• Identify idle spend from over-provisioned infrastructure and workloads.
• Get specific rightsizing recommendations with exact CPU and memory values to apply.
• Prioritise cost reduction without introducing throttling or out-of-memory (OOM) risk.

The view has two tabs: Expenses and Rightsizing.

Expenses

The Expenses tab shows how Kubernetes costs are distributed across your clusters, and what portion of that spend is actively used versus idle.

Summary

At the top of the Expenses tab, four summary cards give a quick read of your Kubernetes spend:

• Total Spend — total cost for the selected date range, split into used and idle components with an efficiency bar.
• CPU Spend — cost attributed to CPU requests across all workloads.
• Memory Spend — cost attributed to memory requests across all workloads.
• Efficiency — the percentage of total spend attributed to active workload usage. A low efficiency percentage indicates a high proportion of idle cost.

Group By

Use the Group By control to change how costs are aggregated in the chart and table. The available grouping dimensions are:

• Namespace (default)
• Cluster
• Node
• Deployment
• Pod
• DaemonSet
• ReplicaSet

Switch the grouping to move from a broad cluster-level view into workload-level or node-level cost attribution.

Cost Breakdown Table

The table groups rows by cluster and shows a cost breakdown for each resource in the selected group-by dimension.

Use the Export CSV button above the table to download the current view for external analysis.

Rightsizing

The Rightsizing tab analyses recent resource usage across your workloads and produces specific recommendations for adjusting CPU and memory requests. Recommendations target over-provisioned workloads for cost reduction, and flag under-provisioned workloads where throttling or OOM events are a risk.

Summary

Four summary cards show the state of your workloads:

• Workloads Scanned — total number of workloads analysed in the selected interval.
• To Shrink — count of workloads with at least one over-provisioned resource (CPU or memory requests can be safely reduced).
• Under-provisioned — count of workloads at risk of CPU throttling or out-of-memory errors.
• Right-sized — count of workloads where no action is needed.

Analysis Interval

The interval selector controls how far back usage metrics are averaged when generating recommendations. Shorter intervals reflect recent behaviour; longer intervals smooth out temporary spikes.

Available intervals: 1H, 3H, 6H, 1D, 1W, 1M.

Choose a longer interval (1W or 1M) when workloads have variable traffic patterns, to avoid rightsizing based on a quiet period. Use a shorter interval when investigating a recent change.

Recommendations Table

Each row in the table represents a workload, showing current resource requests alongside the recommended values.

Decrease rows represent savings opportunities — the workload is using significantly less than it requests, and the requests can be reduced without affecting performance.

Increase rows represent reliability risks — average usage is close to or at the current limit, raising the likelihood of CPU throttling or memory OOM kills. These should be addressed before optimising for cost.

The table sorts by action priority (Decrease first, then Increase, then OK) so the most actionable recommendations appear at the top.

Group By

The Rightsizing tab supports the same grouping dimensions as the Expenses tab: Namespace, Cluster, Node, Deployment (default), Pod, DaemonSet, and ReplicaSet.