Reviewed by Aditya Kumar · Last reviewed 2026-03-25
**Partitioning**: repartition by key before shuffle; coalesce before write. Reduces skew and small files. **Broadcast Join**: Small table < 100MB to all executors; no shuffle. Critical for star schema. **Predicate Pushdown**: Filters pushed to Parquet/ORC; skip row groups....
This hard-level Spark/Big Data question appears frequently in data engineering interviews at companies like Cognizant. While less common, it tests deeper understanding that distinguishes strong candidates. Mastering the underlying concepts (join, optimization, partition) will help you answer variations of this question confidently.
This is a senior-level question that tests architectural thinking. Lead with the high-level design, then drill into specifics. Discuss trade-offs explicitly - there is rarely one correct answer. Show awareness of scale, fault tolerance, and operational complexity.
Partitioning: repartition by key before shuffle; coalesce before write. Reduces skew and small files.
Broadcast Join: Small table < 100MB to all executors; no shuffle. Critical for star schema.
Predicate Pushdown: Filters pushed to Parquet/ORC; skip row groups. Partition pruning on directory structure.
Caching: MEMORY_AND_DISK for reused DFs; unpersist when done.
AQE: Runtime coalesce of shuffle partitions; skew join split; broadcast join switch. Enable spark.sql.adaptive.enabled=true.
Salting: Add random suffix to skewed keys; distribute load; merge after.
Tungsten/Catalyst: Codegen, off-heap, logical optimization.
Scalability Trade-offs: Over-caching evicts other data. Salting adds extra shuffle. AQE adds slight overhead; benefit usually outweighs.
Cost Implications: 2–5x speedup typical with tuning. Profile first—optimize bottleneck, not everything.
Pro-Move: 'EXPLAIN EXTENDED before/after; we validate predicate pushdown in plan.' Red Flag: Tuning without Spark UI—optimizing wrong stage.
Some links below are affiliate links. If you buy through them we may earn a small commission at no extra cost to you — it helps keep DataEngPrep free.
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.