Amazon EKS — VPA — Vertical Pod Autoscaler

Vertical Pod Autoscaler (VPA)

VPA recommends or automatically adjusts CPU and memory requests/limits for pods based on actual usage.

---

VPA Modes

Mode	Behavior
Off	Only generates recommendations (no changes)
Initial	Sets resources at pod creation only
Auto	Evicts and recreates pods with new resources

---

Install VPA

$ git clone https://github.com/kubernetes/autoscaler.git
$ cd autoscaler/vertical-pod-autoscaler
$ ./hack/vpa-up.sh

---

VPA Configuration

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: web-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web
  updatePolicy:
    updateMode: "Off"     # Recommendation-only (safest to start)
  resourcePolicy:
    containerPolicies:
      - containerName: web
        minAllowed:
          cpu: 50m
          memory: 64Mi
        maxAllowed:
          cpu: 2
          memory: 2Gi

---

Check Recommendations

$ kubectl describe vpa web-vpa
  Recommendation:
    Container Recommendations:
      Container Name: web
      Lower Bound:
        Cpu:     25m
        Memory:  128Mi
      Target:
        Cpu:     100m
        Memory:  256Mi
      Upper Bound:
        Cpu:     500m
        Memory:  512Mi

Best practice: Start with Off mode, review recommendations, then apply manually or switch to Auto.

---

HPA + VPA

Approach	Works?
HPA (CPU) + VPA (memory)	✅ Yes — different metrics
HPA (CPU) + VPA (CPU)	❌ No — they’ll conflict

---

Common Mistakes

Mistake	Impact
VPA Auto mode + HPA on same metric	They fight each other
No min/max bounds	VPA might set absurd values
Starting with Auto mode	Unexpected pod restarts

---

What’s Next?

Scale the nodes: Cluster Autoscaler.