Interactive dashboard queries have become slow since a nightly transformation job was moved onto the same engine. What is the appropriate remedy?
Select an answer to reveal the explanation.
Short Explanation
The slowdown is compute contention, so give the batch job its own engine and keep the dashboard on a low-latency engine reading the same tables. You're splitting the compute, not the data.
Full Explanation
The appropriate remedy is to give the batch job its own engine sized for it, leaving the interactive workload on an engine tuned for low latency against the same tables. The cause is two dissimilar workloads sharing one engine. Separating the compute removes the contention while both continue to read the same governed tables, so no copy and no synchronisation is needed.
Copying the dashboard's tables into a separate catalog so the batch job cannot reach them is incorrect because it solves contention by duplicating storage and creating drift, when separating compute achieves it cleanly.
Restricting the dashboard users' permissions so fewer queries are submitted is incorrect because it addresses the symptom by removing legitimate work.
Moving the batch job to run during business hours to spread load evenly is incorrect because it increases overlap with the interactive workload, making the problem worse.
For the exam, respond to contention by separating engines, not by duplicating data or curtailing users.