DP-203 Z-ordering Practice Question
You are optimizing the performance of a large-scale batch processing job in Azure Databricks. The job reads data from Azure Data Lake Storage Gen2, performs transformations, and writes results back. You notice that the job is I/O bound. Which THREE strategies can improve performance? (Choose three.)
⚠ Common exam trap
A common trap is thinking that reducing shuffle partitions (Option D) or enabling autoscaling (Option E) directly address I/O bottlenecks. However, shuffle partitions affect shuffle performance, not storage I/O, and autoscaling adds compute resources, not storage I/O bandwidth. Candidates may also incorrectly believe that caching is only beneficial for compute-bound jobs, but it also reduces I/O.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Use Delta Lake format and optimize the table with Z-ordering on frequently filtered columns.
Option A is correct because Delta Lake with Z-ordering co-locates related data in the same set of files, enabling data-skipping so the I/O-bound job reads far fewer bytes when filtering on those columns. Option B is correct because increasing DataFrame partitions spreads the read/transform workload across more concurrent tasks, raising parallelism and better saturating the storage throughput available to the cluster. Option C is correct because caching the DataFrame in memory (or Delta cache) prevents repeated reads from ADLS Gen2, directly cutting the disk I/O that is the bottleneck. Option D is not correct because reducing shuffle partitions lowers parallelism for wide transformations and typically increases per-task data volume, which does not relieve I/O-bound reads. Option E is not correct because autoscaling adds compute nodes but does not by itself reduce the I/O volume or improve data layout, so it is not a targeted fix for an I/O-bound workload.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Use Delta Lake format and optimize the table with Z-ordering on frequently filtered columns.
Why this is correct
Delta Lake with Z-ordering co-locates related data in the same files using multi-dimensional clustering on frequently filtered columns, enabling data skipping that reads far fewer files. This directly reduces I/O against Azure Data Lake Storage Gen2, satisfying the stem's I/O-bound constraint.
- ✓
Increase the number of partitions in the DataFrame to improve parallelism.
Why this is correct
Increasing DataFrame partitions spreads reads across more Spark tasks, raising parallelism so more executors read from Azure Data Lake Storage Gen2 concurrently. This directly addresses the I/O-bound bottleneck by improving read throughput, satisfying the stem's requirement for a performance strategy.
- ✓
Cache the DataFrame in memory after reading to avoid re-reading from disk.
Why this is correct
Caching the DataFrame in memory with persist or cache keeps transformed data on the cluster, so subsequent actions avoid re-reading from Azure Data Lake Storage Gen2. This directly reduces the disk I/O that makes the batch job I/O bound, satisfying the stem's stated bottleneck.
- ✗
Reduce the number of shuffle partitions to minimize data movement.
Why it's wrong here
Fewer shuffle partitions concentrate work into larger tasks, worsening skew and spilling when data volumes are high. It is tempting because small partition counts reduce task overhead on tiny datasets, but I/O-bound jobs benefit from more partitions, plus file compaction and caching.
- ✗
Enable autoscaling on the cluster to add more nodes during processing.
Why it's wrong here
Autoscaling adds nodes when the cluster is CPU- or memory-bound, but the bottleneck here is storage throughput, so extra workers sit idle waiting on Data Lake reads. It would help a compute-heavy shuffle or aggregation, not an I/O-bound scan of ADLS Gen2.
Go deeper
Related to this question
About these practice questions
One of 509 original DP-203 practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This DP-203 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the DP-203 exam.