Waste Analysis
The Waste Analysis page gives you a category-by-category breakdown of where your Kubernetes clusters are wasting money and how much you could save.
What It Does
K8Cost evaluates every workload in your cluster against 65+ optimization rules across 13 rulesets. The waste analysis page aggregates the results into clear categories so you can prioritize the highest-impact areas first.
Waste Categories
| Category | What It Detects |
|---|---|
| CPU over-provisioning | Pods requesting significantly more CPU than they use (p95 usage vs. request) |
| Memory over-provisioning | Pods with memory requests 1.5x or more above peak usage |
| Idle workloads | Deployments with near-zero CPU and memory utilization for extended periods |
| Unused storage | PVCs that are not mounted by any running pod |
| Over-provisioned storage | PVCs using less than 20% of provisioned capacity |
| Replica waste | Deployments with more replicas than traffic patterns require |
| HPA misconfiguration | Autoscalers with minimum replicas set higher than necessary, or that never trigger a scale event |
| Staging always-on | Non-production workloads running 24/7 that could be scheduled |
Key Capabilities
- Dollar-value breakdown per waste category, so you can see exactly where the money goes
- Trend comparison showing whether waste is growing or shrinking over time
- Namespace-level drill-down to identify which teams or services contribute the most waste
- Severity indicators (critical, warning, info) so you know which issues need immediate attention versus long-term optimization
- Direct link to recommendations -- click any waste category to see the specific recommendations that address it
How Rules Work
Each rule in the K8Cost engine defines a threshold, a severity level, and a remediation template. For example, the CPU over-provisioning rule triggers when a pod's CPU request is more than 2x its p95 usage. The rule generates a recommendation with the suggested new value and an estimated savings figure based on your cloud provider's pricing.
Rules are organized into 13 rulesets covering CPU/memory, idle workloads, replicas, storage, HPA optimization, security configuration, performance, networking, namespace quotas, node utilization, autoscaling, QoS, and AWS-specific cost patterns.
Sign up free to see your cluster's waste breakdown.