Overview

The KEDA add-on scaler provides a mechanism for KEDA add-ons to manage scaling without requiring users to configure complex external scaler details. Instead of manually specifying scalerAddress, metadata, and other configuration, the add-on scaler uses a discovery mechanism where add-ons expose their scaling capabilities through a Custom Resource Definition (CRD).

This approach offers several benefits:

• Automatic Configuration Discovery: Add-ons expose their scaling logic through a CRD status, reducing the need for manual configuration
• Simplified User Experience: Users only need to reference an add-on CRD by name and namespace, not internal scaler details
• Centralized Configuration: All add-on scaling configuration lives in one place (the add-on CRD)

NOTE: This scaler is still experiemtnal

Trigger Specification

This specification describes the keda-add-on trigger for KEDA add-ons.

triggers:
- type: keda-add-on
  metadata:
    name: my-add-on-cr
    kind: MyAddOnKind
    apiVersion: my-api-group/v1alpha1

Parameter list:

• name - The name of the add-on Custom Resource to reference. (Required)
• kind - The kind of the add-on Custom Resource (e.g., InterceptorRoute, HTTPScaledObject). (Required)
• apiVersion - The API version of the add-on Custom Resource (e.g., http.keda.sh/v1alpha1). (Required)

How It Works

Discovery Mechanism

The KEDA add-on scaler follows this process:

1. Fetch CRD: The scaler looks up the referenced Custom Resource by name, namespace, kind, and apiVersion when the ScaledObject|ScaledJob is created or updated
2. Read Status: The scaler reads the CRD’s status field to discover:
   • The external scaler addres
   • Scaling metrics information
   • Default thresholds and target values
3. Delegate Scaling: The scaler uses the discovered information to delegate actual scaling decisions to the add-on’s external scaler (via gRPC)