KEDA — AMP External Scaler Contribution

Fork of kedacore/keda for CNCF upstream contribution.

Contribution: Amazon Managed Prometheus (AMP) External Scaler

Merged PR: kedacore/keda#5315 AWS Blog: Autoscaling Kubernetes workloads with KEDA using Amazon Managed Service for Prometheus metrics

Added a native KEDA scaler for Amazon Managed Prometheus — enables Kubernetes workloads to autoscale on AMP metrics without running self-managed Prometheus.

What was built

• aws_managed_prometheus scaler with AWS SigV4 authentication
• Pod identity (IRSA) support for EKS deployments
• End-to-end tests validating scaler behavior
• AWS SDK v2 integration

Why it matters

Before this scaler, teams using AMP on EKS had to run a self-managed Prometheus instance just to feed metrics to KEDA — defeating the purpose of a managed service. This scaler lets KEDA query AMP directly via IAM, removing the intermediate hop.

Real-world use

The AMP scaler powers event-driven autoscaling in the multi-agent-workflow platform — GPU inference workloads scale on queue depth metrics stored in AMP.

---

For the full KEDA project, see kedacore/keda.