Budgets & Alerts
Set monthly cost budgets for clusters and namespaces, and get notified when spending approaches or exceeds your thresholds.
What It Does
The budget system lets you define monthly spending limits at the cluster or namespace level. K8Cost tracks actual spend against each budget in real time and triggers alerts at configurable thresholds (e.g., 80% and 100% of budget).
Alerts are the notification system that powers budgets, anomaly detection, and recommendation severity escalation. They ensure the right people know about cost issues without having to check the dashboard constantly.
Budget Capabilities
- Cluster-level budgets -- Set a monthly cap for an entire cluster
- Namespace-level budgets -- Set budgets per team or service for granular cost control
- Threshold-based alerts -- Configure warning thresholds (e.g., alert at 50%, 80%, and 100% of budget)
- Trend projection -- K8Cost projects whether current spending will exceed the budget by end of month
- Historical tracking -- See budget utilization over time to calibrate future budgets
Alert Capabilities
- Multiple notification channels -- Email, Slack, Microsoft Teams, PagerDuty, and custom webhooks
- Alert lifecycle -- Alerts progress through active, warning, acknowledged, and resolved states
- Cooldown periods -- Prevent alert fatigue with configurable cooldown between repeat notifications
- Severity levels -- Critical, warning, and informational alerts with different notification urgency
- Bulk acknowledge -- Mark multiple alerts as acknowledged from the dashboard
Common Alert Types
| Alert | Trigger |
|---|---|
| Budget warning | Spend reaches 80% of monthly budget |
| Budget exceeded | Spend exceeds monthly budget |
| Cost anomaly | Spend deviates 2+ standard deviations from baseline |
| Critical recommendation | A critical-severity optimization is detected |
| Agent offline | A cluster agent has not reported within the expected window |
Who Gets Notified
Alerts are delivered based on your notification settings. You can configure different channels for different severity levels -- for example, send warnings to a Slack channel but page on-call via PagerDuty for critical budget breaches.
Sign up free to set up budgets and alerts for your clusters.