In Iceberg, can you explain what a broadcast zone is?

Apache Iceberg is an open‑source table format for huge analytic datasets. A broadcast zone in Iceberg refers to a small, frequently accessed portion of a table that can be efficiently sent (broadcast) to all executors in a cluster, eliminating the need for a shuffle. When a query joins a large table with a small lookup table, Iceberg can mark the small table as a broadcast zone. Spark or Flink will then broadcast the entire small table to each executor, allowing the join to be performed locally. This reduces network traffic, improves latency, and simplifies query planning. Broadcast zones are typically defined by size thresholds (e.g., tables smaller than a few megabytes) or by explicit hints in the query. They are especially useful for dimension tables in star schemas or for lookup tables that change infrequently. By using broadcast zones, Iceberg helps maintain high performance for join-heavy workloads while keeping the underlying data lake architecture scalable.