A company runs a training job that alternates between two phases: one that reads a large dataset sequentially and another that performs random reads on a smaller hot set. The dataset is stored on a remote storage system. The training is I/O bound, and the company wants to optimize performance. Which data placement strategy is most effective?

Use a RAM disk for the large dataset

Cache the hot set on local SSDs and stream the large dataset sequentially

Copy the entire dataset to local SSDs

Use a remote filesystem for both phases

A startup runs a small AI model on a single GPU server. The training data is 50 GB and fits on the server's local SSD. The model training is I/O bound, and the company wants the fastest possible training time. Which data access method should they use?

Stream the dataset from S3 during training

Use a RAM disk to hold the dataset

Access the dataset over NFS from a shared storage server

Copy the dataset to the local SSD before training

A global e-commerce company runs a real-time recommendation model that must respond to user clicks within milliseconds. The model is deployed in a cloud region far from the user base, and the training data is stored in a different region. The company wants to reduce inference latency without moving the model. Which data placement strategy is most effective?

Move the training data to the same region as the model

Use a CDN to cache the model's output

Increase the network bandwidth between the regions

Replicate the model's feature data to the region where the model is deployed

NetAppCertified AI Expert

Domain 5Objective 3

Describe How to Access and Move Data for AI Workloads CERTIFIED-AI-EXPERT Practice Questions (Page 6)

Part of the AI Common Challenges domain, which accounts for 22% of the CERTIFIED-AI-EXPERT exam. NetApp does not publish an official question count, but from its 90-minute exam (~35–60 total, ~8–13 in this domain), expect 1–2 from this objective — we provide 33 practice questions to prepare you well beyond it. (estimate)

33questions here

7free pages

7concepts

22%of the exam

Questions 26–30

26foundation · easy
Which data ingestion pattern is required for AI applications that need to react to data as it is generated, such as fraud detection on live transactions?
Select an answer first
27expert · hard
A company runs a training job that alternates between two phases: one that reads a large dataset sequentially and another that performs random reads on a smaller hot set. The dataset is stored on a remote storage system. The training is I/O bound, and the company wants to optimize performance. Which data placement strategy is most effective?
Select an answer first
28application · medium
A startup runs a small AI model on a single GPU server. The training data is 50 GB and fits on the server's local SSD. The model training is I/O bound, and the company wants the fastest possible training time. Which data access method should they use?
Select an answer first
29application · medium
A global e-commerce company runs a real-time recommendation model that must respond to user clicks within milliseconds. The model is deployed in a cloud region far from the user base, and the training data is stored in a different region. The company wants to reduce inference latency without moving the model. Which data placement strategy is most effective?
Select an answer first
30foundation · easy
Which governance practice ensures that data movement complies with regulations like GDPR by tracking where data originated and where it has been moved?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to CERTIFIED-AI-EXPERT

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by NetApp. “CERTIFIED-AI-EXPERT” is a trademark of its owner, used for identification only.

Describe How to Access and Move Data for AI Workloads CERTIFIED-AI-EXPERT Practice Questions (Page 6)

Questions 26–30

Which data ingestion pattern is required for AI applications that need to react to data as it is generated, such as fraud detection on live transactions?

A startup runs a small AI model on a single GPU server. The training data is 50 GB and fits on the server's local SSD. The model training is I/O bound, and the company wants the fastest possible training time. Which data access method should they use?

Which governance practice ensures that data movement complies with regulations like GDPR by tracking where data originated and where it has been moved?

Keep studying — AI Common Challenges