New feature release: Optimization Policy via CR

Optimization policies can now be managed directly via Kubernetes Custom Resources — alongside or instead of the UI — giving teams full GitOps-style control over how PerfectScale recommends resources across every level of the cluster hierarchy.

By iryna.bohatchenko

Shipped by

Iryna Bohatchenko, Maksym Iemelianov, Abror Kenjaev

What's new:

Multi-level policy control via CRD — set optimization policies (MaxSavings, Balanced, ExtraHeadroom, MaxHeadroom) and custom time windows at cluster, namespace, or workload level using ClusterRecommendationsConfig, NamespaceRecommendationsConfig, and WorkloadRecommendationsConfig CRs
Namespace-level Optimization Policy — for the first time, optimization policies can be configured at the namespace scope, filling the gap between cluster-wide and per-workload settings
Policy hierarchy with smart merging — policies resolve in priority order (Workload CRD → Namespace CRD → Cluster CRD → UI → System Default), so more specific configs always win
CRD source visibility in UI — when a policy is managed via CR, the UI shows its source, displays the raw CR content, and locks editing to prevent conflicts
"Reset inherited" button — workloads with UI-set policies can now be reset to their inherited cluster or namespace default, mirroring the behavior when a CRD is deleted