Kubernetes

CLARITY provides deep visibility into Kubernetes costs across all three major managed services. The Kubernetes page breaks down spending by cluster, namespace, and workload so you can understand exactly where your container costs are going.

Multi-Cluster support

CLARITY discovers and tracks Kubernetes clusters across all connected cloud accounts:

Service	Provider	Discovery
EKS	AWS	Automatic via AWS API
AKS	Azure	Automatic via Azure Resource Manager
GKE	GCP	Automatic via GCP API

All clusters appear in a unified view regardless of provider, with provider badges to distinguish them at a glance. CLARITY is designed to scale: fleets of 60+ clusters stay navigable through grouping, multi-select filters, and a split-view detail panel.

Fleet-scale navigation

Cards vs table view

CLARITY automatically picks the right view for your fleet size:

• Small fleets (1–20 clusters) — default to a card grid showing cost composition, compute mode, and region per cluster.
• Large fleets (20+ clusters) — auto-switch to a dense table with one row per cluster, sticky column headers, and sortable columns for Environment, Mode, Region, Nodes, Control Plane, Compute, Support, Total /mo, Pods, and State.

You can override the auto-switch at any time with the view toggle buttons in the top-right — your choice sticks for the rest of the session.

Split-view detail panel

Clicking any cluster opens a 720px detail sidebar next to the fleet list (not over it). Both panels stay fully interactive: you can click another cluster in the list and the panel updates in place, scroll the fleet table, or use filter controls without closing the drawer. The detail panel is sticky so it follows you as you scroll through a long list.

The detail panel has three tabs:

• Namespace Costs — per-namespace cost breakdown with proportional allocation
• Workload Explorer — deployments, pods, and services inside the cluster
• Label Costs — cost grouped by any Kubernetes label key (team, owner, env, etc.)

Your last-used tab is remembered across clusters, so arrow-key scrubbing through your fleet always lands on the same view.

Grouping & filtering

The toolbar gives you four ways to narrow and organize your fleet:

• Group by — None, Environment, Team, Provider, Region, Compute mode, or Account. Group headers show cluster count and total monthly cost per group, and can be collapsed to focus on specific subsets.
• Multi-select filters — stack Provider + Mode + Environment pills to narrow the list (click to toggle, click All to reset).
• Quick filters:
  • 💰 Top 10 $ — immediately shows the 10 most expensive clusters
  • 🏷️ Untagged — surfaces clusters missing the Environment tag (governance check)
  • ⚠️ With findings — shows only clusters that have open optimization insights
• Search — free-text substring search across cluster name and region

Keyboard shortcuts

Shortcut	Action
/	Focus the search box
↑ / ↓	Navigate through clusters
Esc	Close the detail panel
Ctrl+B / Cmd+B	Toggle the main sidebar

Shortcuts are suppressed while you're typing in any input, so they never interfere with typing a cluster name or email address.

Compute mode awareness

CLARITY automatically detects and displays the compute mode of every cluster so you can see what you're actually paying for:

Mode	Providers	What it means
Managed Nodes	EKS, AKS, GKE Standard	Traditional node pools of EC2 / VM instances you see and size
Fargate	EKS	Serverless per-pod compute — no nodes to manage
Autopilot	GKE	Per-pod billing with Google managing the underlying nodes
Virtual Nodes	AKS	Azure Container Instances (ACI) as a burst backend
Mixed	EKS, AKS	Cluster with multiple compute modes at once (e.g. managed nodes + Fargate profiles)

Each cluster card and table row is provider/mode-aware — the "Compute" column label changes from "Worker Nodes" to "Fargate Tasks" / "Pod Hours" / "VMs + ACI" so you see the right unit for the right cluster.

Cluster-Level cost tracking

Every cluster is broken down into three cost components so you always know where the money is going:

Component	What it is
Control Plane	Fixed per-cluster fee (EKS $2.40/day, AKS Standard $2.40/day, GKE 1st zonal = free, additional zonal = $2.40/day, GKE Autopilot $2.40/day). CLARITY marks free-tier clusters with a badge that explains why.
Worker Compute	Variable cost of nodes, Fargate tasks, Autopilot pod-hours, or Virtual Node ACI containers — whichever applies to the cluster's compute mode.
Support Allocation	Load balancers, persistent volumes, data egress, and other cluster-adjacent costs that don't belong to a single namespace.

The total is displayed as both daily and monthly — in the card view you see all three components stacked with percentage bars, and in the table view they appear as three right-aligned columns plus a bold monthly total.

INFO

Free-tier clusters (AKS Free, GKE 1st zonal) show FREE with a small label explaining the reason so you don't have to remember the provider-specific pricing rules.

Namespace cost allocation

Kubernetes namespaces are the primary unit of cost allocation within a cluster. CLARITY allocates cluster costs to namespaces based on their proportional share of CPU and memory requests.

How it works

1. Total cluster infrastructure cost is calculated from node costs
2. Each namespace's resource requests (CPU + memory) are summed
3. Costs are split proportionally: a namespace requesting 40% of the cluster's total CPU and memory receives 40% of the cost

Metric	Weight
CPU requests	50%
Memory requests	50%

This approach ensures that namespaces consuming more cluster capacity bear a proportionally larger share of the cost.

TIP

Encourage teams to set accurate resource requests on their workloads. Overprovisioned requests inflate the namespace's cost allocation, while missing requests may undercount actual usage.

Compute mode awareness

EKS, AKS, and GKE each ship multiple compute modes. The cluster cards adapt their column labels and cost breakdown so the page tells the truth for each mode:

Mode	Provider(s)	Pricing	Card column label
Managed nodes	EKS / AKS / GKE	Per-instance hours	"Worker Nodes"
Fargate	EKS only	Per-vCPU + per-GB-second	"Fargate Tasks"
Autopilot	GKE only	Per-pod-hour	"Pod Hours"
Virtual nodes	AKS only (rare)	Per-vCPU + per-GB ACI charges	"VMs + ACI"
Mixed	Any	Combination	"Mixed"

The mode is detected automatically from cluster metadata (Fargate-only clusters have fargateOnly=true, GKE Autopilot has isAutopilot=true). Control-plane charges are surfaced separately ($73/mo EKS, free for AKS Standard tier or first GKE zonal cluster, $73/mo GKE Autopilot).

Workload discovery

CLARITY discovers the full workload hierarchy inside each cluster — provider-neutral, the same depth on EKS, AKS, and GKE:

Cluster (eks-cluster / aks-cluster / gke-cluster)
  └─ Namespace (k8s-namespace)
      └─ Deployment / DaemonSet / StatefulSet (k8s-deployment / k8s-daemonset / k8s-statefulset)
          └─ Pod (k8s-pod)

Workload Type	What's Tracked
Namespace	Cost allocation, environment / team tags, total pod count
Deployment / DaemonSet / StatefulSet	Replica count, CPU/memory requests, average utilisation
Pod	Status, restart count, node placement, CPU/memory requests, namespace
Services	Type (ClusterIP, LoadBalancer, NodePort), endpoints

The cluster type carries the provider distinction (eks-cluster, aks-cluster, gke-cluster); workload types beneath are canonical K8s names so the same insight rules and the same UI work uniformly across the three platforms.

K8s-Specific insights

The Insights engine includes rules tailored to Kubernetes environments. Each finding comes with an estimated monthly savings figure, an evidence trail (node count, daily cost, env tag, compute mode), and a concrete next-step recommendation.

Idle clusters

Non-production clusters (sandbox, dev, canary, qa environments) running at full spec 24/7 with minimal workload activity. These are typically forgotten POCs, abandoned migration targets, or demo environments nobody has shut down. CLARITY flags them with severity scaled to monthly cost.

Untagged clusters — governance finding

Clusters with no Owner or Environment tag. Nobody claims responsibility, Finance can't allocate the cost, and the cluster keeps billing. CLARITY surfaces these as a governance finding with the current monthly cost so you can assign an owner or file a termination ticket. Clusters with missing tags also display an inline UNTAGGED pill in the fleet table.

Overprovisioned nodes

Large managed-node clusters (typically >18 nodes) where the aggregate workload requests use a small fraction of available capacity. CLARITY estimates the reclaimable capacity as a percentage of the current compute spend. Consider:

• Reducing the node pool desired count
• Switching to smaller instance types
• Enabling cluster autoscaler with tighter min / max boundaries

Cluster autoscaler disabled

Large production clusters running at fixed size with no autoscaling enabled. Off-peak capacity could be released to save ~18% of compute spend on average. Recommendation includes a suggested min / max node range based on the current count.

Spot/Preemptible opportunities

Non-production clusters running entirely on on-demand nodes. Spot (AWS), Spot (Azure), or Preemptible (GCP) instances can cut compute cost by 60–80% for interruption-tolerant workloads. CLARITY targets this specifically at non-production environments because spot interruption is acceptable there; it does not recommend spot for production prod-tagged clusters without additional context.

Namespace cost outliers

Namespaces whose cost allocation has spiked relative to their historical baseline.

WARNING

Kubernetes insights are validated by the same AI engine as other recommendations. Look for the AI validation badge (Agree / Disagree / Partial) to assess confidence before acting.

Tag governance at fleet scale

Managing tags across dozens of clusters is often where FinOps programs fall apart. CLARITY helps close that gap:

• Untagged column flag — every untagged cluster in the fleet table shows an inline UNTAGGED pill next to its name so you can spot them while scrolling
• Untagged quick filter — one click on the 🏷️ chip narrows the fleet to just the missing-tag clusters
• Governance insights — each untagged cluster over a minimum daily cost threshold generates a finding you can assign to an owner
• Group by Team or Environment — untagged clusters collect in an (untagged) group header so you see the total unowned spend in one number

Node group and pool details

Expand any cluster to see its node groups (AWS), node pools (GCP/Azure), or managed node groups:

• Instance type and pricing tier (on-demand vs. Spot)
• Node count — Current, minimum, and maximum (if autoscaling is enabled)
• Capacity — Total allocatable CPU and memory
• Utilization — Percentage of capacity consumed by scheduled pods

Cost optimization strategies

Common approaches to reduce Kubernetes costs:

Strategy	Potential Savings	Effort
Right-size node instances	15-30%	Medium
Enable cluster autoscaler	10-25%	Low
Use Spot/Preemptible nodes	40-70% on eligible workloads	Medium
Consolidate dev/staging clusters	20-50%	High
Set resource requests accurately	10-20%	Low
Schedule non-prod downtime	30-60% on dev clusters	Medium

What's next?

• Set up namespace-level chargeback
• Review right-sizing recommendations
• Track cluster costs on the Dashboard