How does KEDA enable event-driven autoscaling in Kubernetes, and which custom resource does it use for this purpose?

How does KEDA enable event-driven autoscaling in Kubernetes, and which custom resource does it use for this purpose?

• A. By scaling applications based on external metrics, using ScaledObjects
• B. By deploying a separate operator named ScaledOperator
• C. By creating a standard HorizontalPodAutoscaler
• D. By scaling only on CPU usage through the HPA

Answer: (A)

KEDA enables event-driven autoscaling by letting the system scale workloads in response to external events or metrics, not just how much CPU they use. It does this through a custom resource called a ScaledObject, which links a deployment (or scale target) to one or more triggers that define the external metrics or events driving the scale (like message queue length, pub/sub events, or cloud service metrics) and the desired min and max replicas. The KEDA operator watches these ScaledObject resources and adjusts the scale by translating the external-trigger conditions into the appropriate scale actions within Kubernetes, effectively allowing scaling based on real-world event activity. There isn’t a separate, distinct operator named ScaledOperator, and typical HPA behavior that focuses on CPU utilization is not the primary mechanism here; instead, KEDA uses the ScaledObject to drive scaling from external metrics, while still leveraging Kubernetes underpinnings.

KEDA enables event-driven autoscaling by letting the system scale workloads in response to external events or metrics, not just how much CPU they use. It does this through a custom resource called a ScaledObject, which links a deployment (or scale target) to one or more triggers that define the external metrics or events driving the scale (like message queue length, pub/sub events, or cloud service metrics) and the desired min and max replicas. The KEDA operator watches these ScaledObject resources and adjusts the scale by translating the external-trigger conditions into the appropriate scale actions within Kubernetes, effectively allowing scaling based on real-world event activity. There isn’t a separate, distinct operator named ScaledOperator, and typical HPA behavior that focuses on CPU utilization is not the primary mechanism here; instead, KEDA uses the ScaledObject to drive scaling from external metrics, while still leveraging Kubernetes underpinnings.