An AI training job is running slower than expected. The infrastructure team suspects a compute bottleneck. The job involves processing high-resolution images with a complex convolutional neural network (CNN). They review the logs from the training script.
Epoch 1/100 - 3600s - loss: 1.253 - acc: 0.54
...
CPU_Utilization_During_Epoch: 99% (all cores)
GPU_0_Utilization_During_Epoch: 8%
GPU_1_Utilization_During_Epoch: 7%
...
Epoch 2/100 - 3610s - loss: 1.102 - acc: 0.61
What is the most likely cause of the poor training performance?

• A. The storage system cannot deliver data fast enough to the compute node.
• B. The training script is CPU-bound, likely performing data augmentation or preprocessing on the CPU instead of offloading it to the GPUs.
• C. The network is saturated, preventing the GPUs from receiving data.
• D. The model is too simple and does not effectively utilize the GPU's parallel processing capabilities.

An architect is designing a fully automated, end-to-end MLOps pipeline on Kubernetes for a computer vision use case. The pipeline must handle everything from data versioning to model deployment.
The required pipeline stages are:
1. Data Versioning: Create a new, immutable version of the master dataset for the pipeline run.
2. Data Preparation: Launch a pod to run a preprocessing script on the versioned data.
3. Model Training: Launch a distributed training job that reads the prepared data from a highperformance volume.
4. Model Deployment: Push the trained model to a production inference service.
Which combination of NetApp and Kubernetes technologies provides the most effective and automated solution for this entire pipeline?

• A. Use a single, large ReadWriteMany PVC for all stages to simplify the pipeline configuration.
• B. Use NetApp XCP to copy the data for each stage and configure static PersistentVolumes for each pod.
• C. Use NetApp SnapMirror for data versioning and manually create hostPath volumes for each pipeline stage.
• D. Manually create a NetApp Snapshot via System Manager before each pipeline run, and use the NetApp DataOps Toolkit only for the training stage.
• E. Use the NetApp DataOps Toolkit to create a Snapshot of the source data volume (for versioning), then create a FlexClone PVC from the snapshot for the preparation stage, and finally create a FlexGroup PVC for the training stage.

The pod running the vector database on the Kubernetes cluster fails to start. An MLOps engineer runs 'kubectl describe pod vector-db-pod-0' and sees the following event message:
Events:
Type Reason Age From Message
- - - -
Warning FailedScheduling 30s default-scheduler 0/8 nodes are available: 8 node(s) did not match pod anti-affinity rules.
The pod's manifest contains the following 'affinity' definition:
affinity:
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- key: app
operator: In
values:
- vector-db
topologyKey: "kubernetes.io/hostname"
What is the most likely reason the pod cannot be scheduled?

• A. The 'topologyKey' is invalid; it should be 'failure-domain.beta.kubernetes.io/zone'.
• B. The pod's anti-affinity rule prevents it from being scheduled on any node that is already running another pod with the label 'app=vector-db'.
• C. The PersistentVolumeClaim for the pod is in a 'Pending' state.
• D. The 'vector-db-pod-0' is requesting more GPU resources than are available on any node.

A data scientist needs to launch a Jupyter notebook as a pod in a Kubernetes cluster. The pod requires a 50 Gi persistent volume for storing datasets and notebooks. The cluster administrator has configured a default Trident StorageClass for general-purpose use. The data scientist has the following PersistentVolumeClaim (PVC) manifest:
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: jupyter-pvc
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 50Gi
When this PVC is applied to the cluster, what will be the result?

• A. Trident will automatically provision a 50 Gi volume on its default backend and bind it to the PVC.
• B. Trident will create a 1 Gi volume, as this is the default size for all PVCs.
• C. The PVC will remain in a "Pending" state until a PersistentVolume is manually created.
• D. The PVC will fail because a 'storageClassName' is not explicitly defined.

A data science team works primarily at a central data center but needs to run a short-term, burst- compute training job in the public cloud to take advantage of specialized GPUs that are not available on- premises. They need to efficiently and securely move a 20 TB dataset from their on- premises ONTAP cluster to a Cloud Volumes ONTAP instance for the duration of the job.
The data flow requirement is as follows:
Source: On-premises ONTAP cluster
Destination: Cloud Volumes ONTAP in AWS
Requirement: Efficient, secure, block-level data transfer.
Which NetApp technology is the most appropriate tool for this task?

• A. NetApp FabricPool
• B. NetApp FlexCache
• C. NetApp SnapMirror
• D. NetApp XCP