KubeElasti

Serverless for Kubernetes

Automatically scale your services to zero when idle and scale up when traffic arrives.

KubeElasti saves cost using scale-to-zero without losing any traffic, requires no code changes, and integrates with your existing Kubernetes infrastructure.

Get Started GitHub

Illustration of KubeElasti scale-to-zero lifecycle

Key Features

💰

Cost Optimization

Scale to zero when there's no traffic to save resources and reduce costs

⚡

Zero Downtime

Queues requests during scale-up to ensure no traffic is lost

🔧

Simple Configuration

Easy setup with a single CRD and minimal configuration required

🔄

Seamless Compatibility

Works with your existing Kubernetes setup, HPA, and Keda

📈

Out of Box Monitoring

Built-in monitoring with Prometheus metrics and Grafana dashboards

🛡️

Request Preservation

Ensures all incoming requests are processed even during scale operations

How It Works

Diagram illustrating KubeElasti proxy and serve modes

Scaling Down

When all triggers indicate inactivity, KubeElasti scales the service to 0 replicas and switches to proxy mode

Traffic Queueing

In proxy mode, KubeElasti intercepts and queues incoming requests to the scaled-down service

Scaling Up

When traffic arrives, KubeElasti immediately scales the service back up to its minimum replicas

Serve Mode

Once the service is up, KubeElasti switches to serve mode and processes all queued requests

Serverless with just 1 Command

 
# Create ElastiService CRD for the service you want to optimize
# Replace values between <> with actual values
kubectl apply -f - <<EOF 
apiVersion: elasti.truefoundry.com/v1alpha1
kind: ElastiService
metadata:
  name: <TARGET_SERVICE>
  namespace: <TARGET_SERVICE_NAMESPACE>
spec:
  minTargetReplicas: 1
  service: <TARGET_SERVICE_NAME>
  cooldownPeriod: 5
  scaleTargetRef:
    apiVersion: apps/v1
    kind: deployments
    name: <TARGET_DEPLOYMENT_NAME>
  triggers:
    - type:   prometheus
      metadata:
        # Select a trigger metric to monitor
        query: sum(rate(nginx_ingress_controller_nginx_process_requests_total[1m])) or vector(0)
        # Replace with the address of your Prometheus server
        serverAddress: http://kube-prometheus-stack-prometheus.monitoring.svc.cluster.local:9090
        threshold: "0.5"
EOF

# 🎉 That's it! You just created a scale-to-zero service

KubeElasti is easy to set up and configure. Follow our step-by-step guide to get started.

Full Installation Guide

Join Our Community

Get help, share your experience, and contribute to KubeElasti

🐱 GitHub 🆘 Report Issues 💬 Discussions