Reviewed by Aditya Kumar · Last reviewed 2026-03-25
**ACID**: Transaction log ensures atomic commits; concurrent readers see consistent snapshots. **Time Travel**: versionAsOf, timestampAsOf; audit and rollback. **Schema Enforcement/Evolution**: Reject bad data; add columns additively. **MERGE/UPDATE/DELETE**: Upserts, CDC,...
ACID: Transaction log ensures atomic commits; concurrent readers see consistent snapshots.
Time Travel: versionAsOf, timestampAsOf; audit and rollback.
Schema Enforcement/Evolution: Reject bad data; add columns additively.
MERGE/UPDATE/DELETE: Upserts, CDC, corrections without full overwrite.
OPTIMIZE/Z-Order: Compact files; cluster by columns for read performance.
Unified Batch + Streaming: Same table for batch and streaming sink.
Why These Matter: Traditional lake = files; no transactions, no deletes, no time travel. Delta enables production use cases.
Scalability Trade-offs: Log growth; checkpoint and VACUUM. Z-order write cost.
Cost Implications: Log overhead ~5%; compaction reduces read cost. Enables CDC and late-arriving data handling.
Pro-Move: 'Schema evolution in dev, enforce in prod; mergeSchema for additive only.' Red Flag: Deleting _delta_log manually—corrupts table.
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.