Skip to content

Latest commit

 

History

History
139 lines (95 loc) · 4.73 KB

10-autoscale-my-application.md

File metadata and controls

139 lines (95 loc) · 4.73 KB
Raw

Step 10: Autoscale My Application

Certainly! I'll guide you through the process of implementing Horizontal Pod Autoscaler (HPA) for your e-commerce application. This will allow your application to automatically scale based on CPU usage, helping it handle unpredictable traffic spikes efficiently.

Implement HPA

First, we'll create an HPA targeting 50% CPU utilization, with a minimum of 2 and a maximum of 10 pods. We can do this either by creating a YAML file or using a kubectl command.

Let's create a YAML file named ecommerce-hpa.yaml:

apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
  name: ecommerce-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: ecommerce-website
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      targetAverageUtilization: 50

Apply HPA

Now, let's apply this HPA configuration:

kubectl apply -f ecommerce-hpa.yaml

Alternatively, you can use the kubectl command as you suggested:

kubectl autoscale deployment ecommerce-website --cpu-percent=50 --min=2 --max=10
  1. Verify HPA

Check that the HPA has been created:

kubectl get hpa

You should see output similar to:

NAME            REFERENCE                     TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
ecommerce-hpa   Deployment/ecommerce-website  0%/50%    2         10        2          30s

Simulate Load

To simulate load, we'll use Apache Bench (ab). If you don't have it installed, you can install it on most Unix-based systems with:

sudo apt-get install apache2-utils  # For Ubuntu/Debian
# or
brew install ab  # For macOS with Homebrew

First, get the external IP of your service:

kubectl get service ecommerce-website-service

Now, let's generate some load. Replace <EXTERNAL-IP> with your actual external IP:

ab -n 100000 -c 100 http://<EXTERNAL-IP>/

This command sends 100,000 requests with 100 concurrent connections.

Monitor Autoscaling

While the load is being generated, monitor the HPA:

kubectl get hpa -w

You should see the CPU usage increase and the number of replicas go up:

NAME            REFERENCE                     TARGETS    MINPODS   MAXPODS   REPLICAS   AGE
ecommerce-hpa   Deployment/ecommerce-website  0%/50%     2         10        2          1m
ecommerce-hpa   Deployment/ecommerce-website  60%/50%    2         10        2          2m
ecommerce-hpa   Deployment/ecommerce-website  80%/50%    2         10        3          3m
ecommerce-hpa   Deployment/ecommerce-website  65%/50%    2         10        4          4m
ecommerce-hpa   Deployment/ecommerce-website  55%/50%    2         10        4          5m

You can also watch the pods being created:

kubectl get pods -w
  1. Verify Scaling Down

After the load decreases, the HPA will gradually scale down the number of pods. This may take a few minutes due to the default cooldown period.

Outcome:

The deployment automatically adjusts the number of pods based on CPU load, showcasing Kubernetes' capability to maintain performance under varying loads.

Additional Considerations:

In production, i would consider the following:

  1. Resource Requests: Ensure your deployment specifies CPU requests, as HPA uses these to calculate the target CPU utilization:

    resources:
  requests:
    cpu: 100m

  2. Metrics Server: HPA requires the Metrics Server to be running in your cluster. Most managed Kubernetes services have this pre-installed, but you may need to install it manually on some setups.

  3. Custom Metrics: You can also set up autoscaling based on custom metrics, such as requests per second or queue length, which might be more relevant for your specific application.

  4. Vertical Pod Autoscaler: Consider using Vertical Pod Autoscaler (VPA) in conjunction with HPA to optimize both the number of pods and their resource allocations.

  5. Cluster Autoscaler: If you're running on a cloud provider, consider setting up Cluster Autoscaler to automatically adjust the number of nodes in your cluster based on resource demands.

  6. Cooldown/Delay: Be aware that there's a cooldown period (default is 5 minutes) before scaling down to prevent flapping. You can adjust this if needed.

  7. Monitoring: Set up comprehensive monitoring and alerting to keep track of your application's performance and scaling behavior.

By implementing Horizontal Pod Autoscaler, you've significantly improved your e-commerce application's ability to handle varying loads. This automated scaling ensures that your application remains responsive during traffic spikes while efficiently using resources during periods of low demand.