Correct Answer: Horizontal Pod Autoscaler (HPA) using custom metrics APIs (like KEDA)
While HPA natively supports scaling on CPU/Memory, scaling on custom or external metrics (like Prometheus queue length) requires configuring the custom metrics API or using a frame...
Correct Answer
Horizontal Pod Autoscaler (HPA) using custom metrics APIs (like KEDA)
Why This Answer Is Correct
While HPA natively supports scaling on CPU/Memory, scaling on custom or external metrics (like Prometheus queue length) requires configuring the custom metrics API or using a framework like KEDA (Kubernetes Event-driven Autoscaling) that implements the external metrics provider.
This question evaluates your knowledge of Kubernetes Autoscaling — a core concept tested in cloud certification exams. Understanding Kubernetes Autoscaling is essential for designing scalable, secure, and cost-effective cloud architectures. Mastery of this topic will help you answer scenario-based questions confidently on exam day.
What does a HorizontalPodAutoscaler (HPA) do when CPU utilization exceeds the defined target?
A DevOps team is running a microservice deployment and wants to configure both Horizontal Pod Autoscaler (HPA) to scale out based on CPU usage and Vertical Pod Autoscaler (VPA) to scale resources up/down. What is the recommended practice when using both autoscalers on the same deployment?
During traffic peaks, your Horizontal Pod Autoscaler (HPA) scales out replicas rapidly. However, when traffic dips briefly, the HPA immediately scales down, causing resource thrashing. How can you configure HPA to delay scaling down and stabilize replica count?
You are running a stateful pod that performs long-running computations. You want to ensure that the Cluster Autoscaler (CA) does not terminate the node hosting this specific pod during a cluster scale-down event. Which annotation should you apply to the pod?