Yesterday, the query plan for the nightly reconciliation broadcast a 30 MB currency dimension next to a multi-terabyte fact table automatically — no hint appears in the query. Today the same pattern runs as a shuffle-heavy sort-merge join, because the platform team lowered the size threshold that governs automatic broadcasts of small tables in joins. Which configuration parameter sets that threshold?

1. A. spark.sql.autoBroadcastJoinThreshold — the size limit at or below which the planner broadcasts a join side automatically, with no hint required ✓ A
2. B. No parameter is involved — automatic broadcast joins only happen when the query text includes an explicit broadcast() join hint 💬 B
3. C. The broadcast threshold is a fixed engine constant — a platform team cannot raise or lower it through configuration 💬 C
4. D. spark.sql.shufflePartitions — its value determines how large a table must be before the planner stops broadcasting it 💬 D

Revelar gabarito

Resposta correta: spark.sql.autoBroadcastJoinThreshold — the size limit at or below which the planner broadcasts a join side automatically, with no hint required

Referência: https://docs.databricks.com/aws/en/optimizations/aqe