A Delta‑backed star schema joins one large fact table with several small dimension tables on equality keys. What SQL tactic usually gives the best first‑step speedup?

The fastest first‑step optimization is to broadcast the small dimension tables. Use a BROADCAST hint (or set spark.sql.autoBroadcastJoinThreshold appropriately) so the optimizer pushes the dimensions into memory on each executor. Ensure accurate statistics are available so the planner chooses the broadcast plan. This eliminates shuffle, reduces I/O, and dramatically speeds up the join, especially when the fact table is large and the dimensions are tiny.