Observed arrival · 2026-11-10
Kerf puts guardrails around Kubernetes rightsizing
An open-source tool that uses container usage data to recommend and apply Kubernetes CPU and memory settings, with cost estimates and rollback safeguards.
- For
- Kubernetes platform teams managing multiple clusters
- Worth noticing
- The production profile uses the 95th percentile of 14 days of usage plus 15% headroom, with a confidence score.
Field notes
The page lays out a hub-and-agent topology: the hub computes recommendations while agents in workload clusters apply changes under per-resource policies. Agents pull over HTTPS, so the hub does not connect back into those clusters. The production profile is described as using 14 days of usage data, and savings can be priced by node type using AWS prices or a user-supplied table. The project links to an Apache-2.0 license, documentation, releases, and a changelog.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue