Reviewed by Aditya Kumar · Last reviewed 2026-03-24
**Skew**: Few partitions have most data—stragglers; one task runs 10× longer. **Too many**: Scheduler overhead; small tasks inefficient; many small files. **Too few**: Underutilized cluster; OOM on large partitions. **Wrong key**: Full shuffle; no pruning. **Symptoms**: Uneven...
Skew: Few partitions have most data—stragglers; one task runs 10× longer. Too many: Scheduler overhead; small tasks inefficient; many small files. Too few: Underutilized cluster; OOM on large partitions. Wrong key: Full shuffle; no pruning. Symptoms: Uneven task progress; OOM; long tail. Fix: Repartition; salt skewed keys; align with data distribution and query patterns.
Red Flag: Default 200 partitions without profiling. Pro-Move: 'We profiled: 30% of keys were NULL—salted NULL into 10 buckets; eliminated stragglers, job 4× faster.'
According to DataEngPrep.tech, this is one of the most frequently asked SQL interview questions, reported at 1 company. DataEngPrep.tech maintains an editor-reviewed database of 1,863 data engineering interview questions across 7 categories.