Google Associate Cloud Engineer
Get started today
Ultimate access to all questions.
Your team of data scientists occasionally needs access to a Google Kubernetes Engine (GKE) cluster under your management. They require GPU resources to execute certain lengthy, non-restartable tasks. To manage this while minimizing costs, what approach should you take?
Your team of data scientists occasionally needs access to a Google Kubernetes Engine (GKE) cluster under your management. They require GPU resources to execute certain lengthy, non-restartable tasks. To manage this while minimizing costs, what approach should you take?
Explanation:
The correct answer is A: Enable node auto-provisioning on the GKE cluster. Node auto-provisioning allows for the creation and deletion of node pools based on the current workload requirements. This helps in minimizing costs as it only provisions resources when needed. While option D also allows autoscaling, it will maintain a minimum of one instance at all times, which could result in higher costs because it doesn't scale down to zero when not in use. Options B and C are not suitable because VerticalPodAutoscaler optimizes for Pods and preemptible VMs are not ideal for long-running, non-restartable jobs due to their potential for interruption.