Demonstrate the Ability to Execute AI Workloads at Scale with Kubernetes (Trident) CERTIFIED-AI-EXPERT Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   What is the purpose of setting resource requests and limits for CPU and memory in an AI pod's specification?

   • ARequests and limits are only used for monitoring and have no effect on scheduling or execution.
   • BRequests define the maximum resources a pod can use, while limits define the minimum.
   • CLimits are used for scheduling, while requests are used for runtime enforcement.
   • DRequests guarantee a minimum amount of resources for scheduling, while limits cap the maximum usage to prevent resource contention.
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   Which Kubernetes feature ensures that a specified number of AI workload pods are always running, providing high availability in case of node failures?

   • AService
   • BNamespace
   • CReplicaSet (managed by a Deployment)
   • DConfigMap
   Reveal Solution Discussion Select an answer first
4. 13expert · hard

   A company runs a stateful AI workload that requires persistent storage. They want to ensure that if a pod is rescheduled to a different node, it can still access its data. They also want to protect against data loss by taking regular snapshots. Which combination of Kubernetes and Trident features should they use?

   • AA Deployment with a PVC and a VolumeSnapshot schedule.
   • BA Job with an emptyDir volume and a backup cron job.
   • CA DaemonSet with a hostPath volume and manual snapshots.
   • DA StatefulSet with a volumeClaimTemplate and a VolumeSnapshot schedule.
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   In Kubernetes, which object is the smallest deployable unit that can be created and managed, and typically contains one or more containers that share the same network namespace and storage volumes?

   • AService
   • BDeployment
   • CNamespace
   • DPod
   Reveal Solution Discussion Select an answer first
6. 15expert · hard

   A Kubernetes cluster has a mix of nodes with different CPU architectures (amd64 and arm64). An AI workload is compiled only for amd64. The team wants to ensure that the workload runs only on amd64 nodes. They also want to ensure that the workload gets priority over other workloads on those nodes. What should they configure?

   • AA taint on arm64 nodes and a toleration on the workload.
   • BA resource quota for amd64 nodes.
   • CA nodeAffinity for amd64 and a low-priority PriorityClass.
   • DA nodeSelector for amd64 and a high-priority PriorityClass.
   Reveal Solution Discussion Select an answer first