openova/platform/vpa

Vertical Pod Autoscaler (VPA)

Automated resource right-sizing. Per-host-cluster infrastructure (see docs/ARCHITECTURE.md §3.4) — runs on every host cluster a Sovereign owns.

Status: Accepted | Updated: 2026-04-27

---

Overview

VPA provides automated resource optimization:

• Reduces over-provisioning waste
• Prevents under-provisioning issues
• Works alongside horizontal scaling (KEDA)
• Provides recommendations even if not auto-applying

---

Architecture

flowchart TB
    subgraph VPA["VPA Components"]
        Rec[Recommender]
        Upd[Updater]
        Adm[Admission Controller]
    end

    subgraph Metrics["Metrics"]
        MS[Metrics Server]
        Prom[Prometheus/Mimir]
    end

    subgraph Workloads["Workloads"]
        Deploy[Deployments]
        Pods[Pods]
    end

    MS --> Rec
    Prom --> Rec
    Rec -->|"Recommendations"| Upd
    Upd -->|"Evict pods"| Pods
    Adm -->|"Mutate requests"| Pods
    Deploy --> Pods

---

Update Modes

Mode	Behavior	Use Case
Off	Recommendations only	Analysis, not auto-apply
Initial	Apply on pod creation	Batch jobs
Auto	Evict and recreate	Long-running services
Recreate	Same as Auto	Legacy compatibility

Recommended: Auto for most workloads

---

Configuration

VPA Resource

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: <org>-app-vpa
  namespace: <org>
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: <org>-app
  updatePolicy:
    updateMode: Auto
  resourcePolicy:
    containerPolicies:
      - containerName: "*"
        minAllowed:
          cpu: 50m
          memory: 64Mi
        maxAllowed:
          cpu: 2
          memory: 4Gi
        controlledResources:
          - cpu
          - memory
        controlledValues: RequestsAndLimits

Kyverno Auto-Generation

Kyverno automatically generates VPAs for deployments:

apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata:
  name: generate-vpa
spec:
  rules:
    - name: generate-vpa-for-deployment
      match:
        any:
          - resources:
              kinds:
                - Deployment
      exclude:
        any:
          - resources:
              annotations:
                vpa.openova.io/skip: "true"
      generate:
        apiVersion: autoscaling.k8s.io/v1
        kind: VerticalPodAutoscaler
        name: "{{request.object.metadata.name}}-vpa"
        namespace: "{{request.object.metadata.namespace}}"
        data:
          spec:
            targetRef:
              apiVersion: apps/v1
              kind: Deployment
              name: "{{request.object.metadata.name}}"
            updatePolicy:
              updateMode: Auto

---

VPA + KEDA Interaction

flowchart LR
    subgraph Scaling["Scaling"]
        VPA[VPA<br/>Vertical]
        KEDA[KEDA<br/>Horizontal]
    end

    subgraph Workload["Workload"]
        Deploy[Deployment]
        Pods[Pods]
    end

    VPA -->|"Right-size resources"| Pods
    KEDA -->|"Scale replicas"| Deploy
    Deploy --> Pods

Coordination:

• VPA handles CPU/memory per pod
• KEDA handles replica count
• Combined: optimal resource utilization

---

Monitoring

Metric	Description
vpa_recommender_*	Recommender metrics
vpa_updater_*	Updater metrics
container_resource_recommendations	Per-container recommendations

Dashboard

Grafana dashboard shows:

• Current vs recommended resources
• Historical recommendations
• Eviction events
• Cost savings estimates

---

Part of OpenOva