Reviewed by Aditya Kumar · Last reviewed 2026-03-25
**Narrow**: No shuffle. Each partition processed independently. map, filter, flatMap. Can stay in single partition. **Wide**: Requires shuffle. Data from multiple partitions combined. join, groupBy, distinct, repartition. **Why It Matters**: Narrow = cheap, parallel. Wide =...
Narrow: No shuffle. Each partition processed independently. map, filter, flatMap. Can stay in single partition.
Wide: Requires shuffle. Data from multiple partitions combined. join, groupBy, distinct, repartition.
Why It Matters: Narrow = cheap, parallel. Wide = expensive; stage boundary. DAG split at wide.
Examples: filter (narrow), reduceByKey (wide).
Scalability Trade-offs: Minimize wide. Push filters early (narrow). Design narrow-first.
Cost Implications: Wide = bulk of cost. Reduce wide transformations.
Pro-Move: 'We count wide transformations per job; target <5.' Red Flag: Wide transformation when narrow possible—e.g., filter after join.
According to DataEngPrep.tech, this is one of the most frequently asked Spark/Big Data interview questions, reported at 1 company. DataEngPrep.tech maintains an editor-reviewed database of 1,863 data engineering interview questions across 7 categories.