Cost Optimization
Atmosly provides built-in cost tracking and optimization insights for your Kubernetes clusters, helping you understand and control your cloud spending.
Overview
The Cost Optimization module tracks infrastructure costs at multiple levels — from the overall cluster down to individual node groups and workloads. This visibility helps teams identify over-provisioned resources and optimize spending.
Cost analysis is available for clusters provisioned through Atmosly or imported via cloud account integration. Token-imported clusters do not support cost analysis.
Cluster Cost Overview
Navigate to a cluster's Cost section to view:
| Metric | Description |
|---|---|
| Total Cluster Cost | Aggregate cost of all resources in the cluster |
| Cost Trend | Historical cost data showing spending patterns over time |
| Cost Breakdown | Breakdown by resource type (compute, storage, networking) |
| Node Group Costs | Per-node-group cost allocation |
Cost Tracking Levels
Cluster-Level Costs
- Total cost across all node groups, storage, and networking
- Historical trend analysis
- Cost comparison across time periods
Node Group-Level Costs
- Per-node-group cost breakdown
- Instance type cost analysis
- Spot vs. On-Demand cost comparison
- Idle resource identification
Pod and Node Cost Scanning
- Background cost scanning for individual pods and nodes
- Resource utilization vs. cost efficiency analysis
- Recommendations for right-sizing
Right-Sizing: Reviewing and Applying Fixes
Beyond surfacing recommendations, Atmosly can apply right-sizing changes to your workloads — a closed loop of recommend → preview → apply → measure → revert.
For a workload with a recommendation, open its right-sizing panel to review the suggested CPU/memory request and limit changes.
Preview the change
Before applying anything, preview the diff to see exactly what will change on the workload, and confirm the source Atmosly will modify (the live object vs. the manifest in Git).
Apply options
| Method | What it does | When to use |
|---|---|---|
| Raise a GitOps PR | Opens a pull request against the workload's source manifests with the new values. | Workloads managed via GitOps (e.g., Argo CD) — keeps Git as the source of truth. |
| Apply directly | Applies the change to the live cluster immediately via the ops-agent. | Workloads not managed by GitOps, or when you need the change to take effect right away. |
Revert
An applied fix can be reverted from the same panel for a limited time after it's applied, so you can quickly roll back if the new sizing doesn't behave as expected.
If a workload is reconciled by a GitOps tool such as Argo CD, a direct apply is temporary — the next sync will reconcile the workload back to what's defined in Git. For a permanent change, either raise a GitOps PR (recommended) or update the source manifests yourself. Atmosly can pause auto-sync briefly while a direct fix is applied so it isn't reconciled away immediately.
Applying or reverting a right-sizing recommendation requires the apply recommendation permission.
Cost Estimation (During Cluster Creation)
When creating a new cluster, Atmosly provides a cost estimation preview based on your selected configuration:
- Instance types and count
- Storage allocation
- Add-on overhead
- Estimated monthly cost
This helps you make informed decisions about cluster sizing before provisioning.
Supported Cloud Providers
| Provider | Cost Tracking | Cost Estimation |
|---|---|---|
| AWS (EKS) | Yes | Yes |
| GCP (GKE) | Yes | Yes |
| Token Import | No | No |