VPA Memory Lab

Welcome to the VPA Memory Lab! In this tutorial, you’ll deploy a sample Flask application on Kubernetes, monitor its memory usage, configure a Vertical Pod Autoscaler (VPA), generate load, and review VPA memory recommendations. By the end, you’ll understand how VPA adjusts resource requests to match real-world demand.

Step 1: Deploy the Sample Flask Application

Apply the deployment and service manifest:

kubectl apply -f vpa-testing.yml

Expected output:

deployment.apps/flask-app created
service/flask-app-service created

Confirm the pods are running:

kubectl get pods

NAME                        READY   STATUS    RESTARTS   AGE
flask-app-b85fc57d4-kmsdl   1/1     Running   0          2m
flask-app-b85fc57d4-zn77q   1/1     Running   0          2m

Step 2: Check Current Resource Usage

Before load testing, inspect CPU and memory usage:

kubectl top pod

NAME                        CPU(cores)   MEMORY(bytes)
flask-app-b85fc57d4-g2n7n   1m           19Mi
flask-app-b85fc57d4-mq6kb   1m           19Mi

Step 3: Apply the VPA Configuration

Create a VPA manifest (vpa-memory.yaml) to manage memory requests for your Flask deployment:

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: flask-app
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: flask-app
  updatePolicy:
    updateMode: "Off"
  resourcePolicy:
    containerPolicies:
      - containerName: '*'
        controlledResources: ["memory"]
        minAllowed:
          memory: 150Mi
        maxAllowed:
          memory: 1000Mi

Apply the VPA manifest:

kubectl apply -f vpa-memory.yaml

verticalpodautoscaler.autoscaling.k8s.io/flask-app created

Initial VPA Recommendation

Check the initial memory recommendation:

kubectl get vpa flask-app -o yaml

status:
  conditions:
    - type: RecommendationProvided
      status: "True"
      lastTransitionTime: "2025-01-15T08:10:03Z"
  recommendation:
    containerRecommendations:
      - containerName: flask-app
        lowerBound:
          memory: 262144k
        target:
          memory: 262144k
        uncappedTarget:
          memory: 262144k
        upperBound:
          memory: 1000Mi

Step 4: Run a Load Test and Validate Recommendations

Generate load against your Flask application:

sh load.sh

Once the load test completes, retrieve the updated VPA recommendation:

kubectl get vpa flask-app -o yaml

status:
  recommendation:
    containerRecommendations:
      - containerName: flask-app
        lowerBound:
          memory: 262144k
        target:
          memory: "511772K"
        uncappedTarget:
          memory: "511772K"
        upperBound:
          memory: 1000Mi

Notice that target has increased to meet higher memory demands under load. If the uncappedTarget exceeds your upperBound, you can:

• Adjust maxAllowed in the VPA policy
• Consider a Horizontal Pod Autoscaler for scaling out

Links and References

• Kubernetes Vertical Pod Autoscaler
• VPA GitHub Repository
• Kubernetes Autoscaling Documentation