Preparing Practice Exam…
Please wait a moment while the system generates the exam questions, answers, and study session.
Please wait a moment while the system generates the exam questions, answers, and study session.
Databricks Certified Data Engineer Professional · Databricks
Select one answer.
Written by other learners. Comments can disagree with the marked answer — check the official documentation when in doubt.
Going for A, it still has same cores but no need to shuffle across the network.
The answer is A, and the "more executors = more parallelism" reasoning doesn't hold here. Total cores are fixed at 160 in every option. Parallelism in Spark is driven by total executor cores, not by executor count: A gives 1 × 160 = 160 concurrent tasks, C gives 16 × 10 = 160. Identical. The Databricks docs make this explicit — total executor cores determine max parallelism, and 2 workers × 16 cores is the same compute as 8 workers × 4 cores. Since parallelism is a wash, the only differentiator is shuffle cost. One executor = shuffle stays local, zero network I/O and zero serialization across nodes. 16 nodes means every wide transformation crosses the wire.
Changed my mind. The reason is that we have to focus on best distributed task among executors and option C will give more parallelism