問題文
A team upgrades to a recent Apache Spark release and hears that the engine can revise a query plan while the query runs. Which statement about adaptive query execution is correct?
選択肢
- It uses statistics collected in advance by an analyze command, and it must be enabled with a configuration flag before any of its behaviors take effect on a query that has already been planned by the optimizer.
- It replaces the cost-based optimizer entirely, so table statistics no longer matter, and the analyze command becomes a formality.
- It only reorders joins, and it does not change the number of partitions, so a shuffle keeps the same task count throughout the query.
- It uses statistics gathered at run time to revise the plan, and it is enabled by default in recent releases.