Horizontal Pod Autoscaler Pack

Pod Autoscaling That Actually Works — CPU, Memory, and Queue Depth

HPA is straightforward until you need to scale on something other than CPU — then you need KEDA, a ScaledObject, a trigger, and a test to prove it works. This pack covers all three scaling patterns with working configurations and load tests to validate each one.

What is Included

• HPA manifests for CPU-based and memory-based scaling with tuned stabilization windows
• KEDA ScaledObject configurations for CPU, memory, and SQS queue depth triggers
• KEDA IRSA policy for SQS queue length polling
• Load testing scripts (k6) to drive CPU and memory pressure and verify scale-up events
• SQS load generator script to validate queue-depth scaling

Developer Experience

Each scaling pattern is self-contained with its own manifest and test script. Run the test, watch the pods scale, confirm the behavior. Stabilization window and cooldown settings are annotated so you understand the tradeoffs.

What You Get

• hpa/ — CPU and memory HPA manifests
• keda/ — ScaledObject configs for CPU, memory, and SQS
• tests/ — k6 load scripts and SQS message generator
• README.md covering KEDA install, IRSA setup, and expected scaling timeline

Requirements

• Kubernetes >= 1.23 with metrics-server installed
• KEDA >= 2.12 (install guide included)
• AWS SQS queue for queue-depth scaling (optional pattern)