KEDA Integration

KEDA (Kubernetes Event Driven Autoscaler) is the recommended way to trigger Dequeueable in a Kubernetes cluster. KEDA monitors your Azure Queue Storage and spins up a new job instance for each message, shutting it down when the queue is empty.

How It Works

KEDA uses a ScaledJob resource to manage job scaling. Unlike a ScaledObject (which scales a Deployment), a ScaledJob creates a new Kubernetes Job for each trigger — which maps perfectly to Dequeueable's ephemeral model.

Queue has 5 messages → KEDA creates 5 Job instances → each processes 1 message → all shut down

Setup

Follow the official KEDA documentation to get started:

• Installing KEDA
• Azure Queue Storage Scaler
• ScaledJob

Recommendations

When configuring the Azure Queue Storage scaler, set queueLength to 1 to ensure KEDA creates exactly one job instance per message, matching Dequeueable's single-message model.

Since each job instance processes exactly one message, you can set tight and predictable resource limits on your container. This is one of the key advantages over a shared host model — resource usage per message is consistent and enforceable at the container level.

For production, use Azure Workload Identity to authenticate KEDA with your storage account without storing credentials in Kubernetes secrets.